Update Space And Models

Authorizations

Authorization

string

header

required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json

string

required

app_logo

string | null

benchmark_model_id

string | null

Used to update the benchmark model

description

string | null

has_failover_chain

boolean | null

name

string | null

Minimum length: 1

policies

object | null

policies.max_cost

number

default:

0.02

The maximum cost allowed for a request. Only works with compounded requests that require multiple LLM calls. If the value is reached, it will exit with an exception.

Required range: x > 0.0001

policies.max_same_model_retries

integer

default:

If an LLM call fails, how many times should Pulze retry the call to the same LLM? There will be a maximum of N+1 calls (original + N retries)

Required range: 0 < x < 3

policies.max_switch_model_retries

integer

default:

If an LLM call fails, how many other models should Pulze try, chosen by quality descending? It will be a maximum of N+1 models (original + N other models)

Required range: 0 < x < 5

policies.optimize_internal_requests

enum<integer>

default:

Optimize the internal / intermediate LLM requests, for a big gain in speed and cost savings, at the cost of a potential, and very slight, penalty on quality. The final request ("SYNTHESIZE") is always performed using your original settings.

Available options:

0,

1

policies.privacy_level

enum<integer>

default:

    The level of privacy for a given request
    0 = (UNSUPPORTED -- public logs)
    1 = Log request, response and all of its metadata (Normal mode)
    2 = Do not log neither the request prompt nor the response text. Logs are still visible, and all of the request metadata accessible. Retrievable as a log. (TBD)
    3 = Do not log at all. Internally, a minimal representation may be stored for billing: model name, tokens used, which app it belongs to, and timestamp. Not retrievable as a log. (TBD)

Available options:

1,

2,

3

policies.prompt_id

string | null

Prompt ID that we will use for requests

sandbox_of

string | null

The app this sandbox relates to

settings

object | null

use_parent_custom_data

boolean | null

weights

object | null

Response

200 - application/json

main

object

required

main.active_models

integer

required

main.app

object

required

main.app.auth0_id

string

required

The Auth0ID of the creator of the app

main.app.benchmark_model

object

required

Compare the results of LLMs against this model (for speed, quality, cost...)

main.app.benchmark_model.added_by

string | null

required

The user (auth0_id) who created the model

main.app.benchmark_model.added_on

string | null

required

When the model was added. Auto-populated in DB

main.app.benchmark_model.app_id

string | null

required

The app_id that has access to this model (if only one)

main.app.benchmark_model.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

main.app.benchmark_model.completion_token_cost

number

required

The cost of a completion token, in USD

main.app.benchmark_model.context_window

integer

required

The max_tokens for this model

main.app.benchmark_model.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

main.app.benchmark_model.description

string

required

A description of the model

main.app.benchmark_model.id

string | null

required

The ID of this model

main.app.benchmark_model.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

main.app.benchmark_model.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

main.app.benchmark_model.is_open_source

boolean

required

True if the model is open source

main.app.benchmark_model.is_public

boolean

required

True if the model is publicly accessible to all

main.app.benchmark_model.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

main.app.benchmark_model.is_rag

boolean

required

Whether it's rag-tuned or not

main.app.benchmark_model.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

main.app.benchmark_model.model

string

required

The name of the model. Can belong to many providers

main.app.benchmark_model.modified_on

string | null

required

When the model was updated. Auto-populated in DB

main.app.benchmark_model.namespace

string

required

The fully qualified (namespaced) model name

main.app.benchmark_model.org_id

string | null

required

The org_id that has acccess to this model

main.app.benchmark_model.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

main.app.benchmark_model.prompt_token_cost

number

required

The cost of a prompt token, in USD

main.app.benchmark_model.supports_functions

boolean

required

True if the model supports function/tool call

main.app.benchmark_model.supports_json

boolean

required

True if the model supports json-formatted responses

main.app.benchmark_model.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

main.app.benchmark_model.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

main.app.benchmark_model.supports_stream

boolean

required

True if the model supports streaming responses

main.app.benchmark_model.supports_vision

boolean

required

True if the model supports image recognition (vision)

main.app.benchmark_model.until

string | null

required

The most recent data this model has been trained with

main.app.benchmark_model.url

string

required

A URL to the model's page or more informatino

main.app.benchmark_model.api_target

string

default:

Store the name of the model the API requires

main.app.benchmark_model.at

string | null

Extra model settings inferred from namespace

main.app.benchmark_model.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.app.benchmark_model.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.app.benchmark_model.parent

object | null

main.app.benchmark_model.parent.added_by

string | null

required

The user (auth0_id) who created the model

main.app.benchmark_model.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

main.app.benchmark_model.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

main.app.benchmark_model.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

main.app.benchmark_model.parent.completion_token_cost

number

required

The cost of a completion token, in USD

main.app.benchmark_model.parent.context_window

integer

required

The max_tokens for this model

main.app.benchmark_model.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

main.app.benchmark_model.parent.description

string

required

A description of the model

main.app.benchmark_model.parent.id

string | null

required

The ID of this model

main.app.benchmark_model.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

main.app.benchmark_model.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

main.app.benchmark_model.parent.is_open_source

boolean

required

True if the model is open source

main.app.benchmark_model.parent.is_public

boolean

required

True if the model is publicly accessible to all

main.app.benchmark_model.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

main.app.benchmark_model.parent.is_rag

boolean

required

Whether it's rag-tuned or not

main.app.benchmark_model.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

main.app.benchmark_model.parent.model

string

required

The name of the model. Can belong to many providers

main.app.benchmark_model.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

main.app.benchmark_model.parent.namespace

string

required

The fully qualified (namespaced) model name

main.app.benchmark_model.parent.org_id

string | null

required

The org_id that has acccess to this model

main.app.benchmark_model.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

main.app.benchmark_model.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

main.app.benchmark_model.parent.supports_functions

boolean

required

True if the model supports function/tool call

main.app.benchmark_model.parent.supports_json

boolean

required

True if the model supports json-formatted responses

main.app.benchmark_model.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

main.app.benchmark_model.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

main.app.benchmark_model.parent.supports_stream

boolean

required

True if the model supports streaming responses

main.app.benchmark_model.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

main.app.benchmark_model.parent.until

string | null

required

The most recent data this model has been trained with

main.app.benchmark_model.parent.url

string

required

A URL to the model's page or more informatino

main.app.benchmark_model.parent.api_target

string

default:

Store the name of the model the API requires

main.app.benchmark_model.parent.at

string | null

Extra model settings inferred from namespace

main.app.benchmark_model.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.app.benchmark_model.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.app.benchmark_model.parent.parent

object | null

main.app.benchmark_model.parent.parent_id

string | null

The ID of parent, in case it's not a base model

main.app.benchmark_model.parent.prompt_id

string | null

The ID of prompt, used for this model

main.app.benchmark_model.parent.provider

string | null

The provider for the model.

main.app.benchmark_model.parent_id

string | null

The ID of parent, in case it's not a base model

main.app.benchmark_model.prompt_id

string | null

The ID of prompt, used for this model

main.app.benchmark_model.provider

string | null

The provider for the model.

main.app.benchmark_model_id

string

required

Compare the results of LLMs against this model (for speed, quality, cost...)

main.app.has_failover_chain

boolean

required

A Failover chain, when enabled, skips the SMART router and instead calls the failover chain in order.

main.app.hashed_key

string

required

The hashed value of the API Key which we store in our database

main.app.id

string

required

main.app.is_active

boolean

required

True if the app is active (not soft-deleted, etc.), false otherwise

main.app.key_end

string

required

The last characters of the API key

main.app.modified_by

string

required

The Auth0ID of the last person to modify the table

main.app.name

string

required

A name for this app

main.app.prompt

object | null

required

The Prompt associated to the app, if any

main.app.added_on

string

main.app.app_logo

string | null

A logo for this app

main.app.description

string | null

A description for this app

main.app.modified_on

string

main.app.org_id

string

The Org to which this App belongs

main.app.policies

object

main.app.policies.max_cost

number

default:

0.02

The maximum cost allowed for a request. Only works with compounded requests that require multiple LLM calls. If the value is reached, it will exit with an exception.

Required range: x > 0.0001

main.app.policies.max_same_model_retries

integer

default:

If an LLM call fails, how many times should Pulze retry the call to the same LLM? There will be a maximum of N+1 calls (original + N retries)

Required range: 0 < x < 3

main.app.policies.max_switch_model_retries

integer

default:

If an LLM call fails, how many other models should Pulze try, chosen by quality descending? It will be a maximum of N+1 models (original + N other models)

Required range: 0 < x < 5

main.app.policies.optimize_internal_requests

enum<integer>

default:

Available options:

0,

1

main.app.policies.privacy_level

enum<integer>

default:

    The level of privacy for a given request
    0 = (UNSUPPORTED -- public logs)
    1 = Log request, response and all of its metadata (Normal mode)
    2 = Do not log neither the request prompt nor the response text. Logs are still visible, and all of the request metadata accessible. Retrievable as a log. (TBD)
    3 = Do not log at all. Internally, a minimal representation may be stored for billing: model name, tokens used, which app it belongs to, and timestamp. Not retrievable as a log. (TBD)

Available options:

1,

2,

3

main.app.policies.prompt_id

string | null

Prompt ID that we will use for requests

main.app.prompt_id

string | null

If any, the ID of the prompt associated with this app

main.app.rate_limit

integer

default:

200

main.app.sandbox_of

string | null

The app might be a (testing) sandbox of a different App.

main.app.settings

object

main.app.use_parent_custom_data

boolean | null

If true, the app will use the custom data of the parent app

main.app.weights

object

main.base_model_settings

object[]

required

main.base_model_settings.context_window

integer

required

The max_tokens for this model

main.base_model_settings.description

string

required

A description of the model

main.base_model_settings.id

string | null

required

Used to uniquely target models when we enable/disable them

main.base_model_settings.model

string

required

The name of the model. Can belong to many providers

main.base_model_settings.namespace

string

required

The fully qualified (namespaced) model name

main.base_model_settings.until

string | null

required

The most recent data this model has been trained with

main.base_model_settings.url

string

required

A URL to the model's page or more informatino

main.base_model_settings.active

boolean

default:

false

Whether the model is active for the app.

main.base_model_settings.api_target

string

default:

Store the name of the model the API requires

main.base_model_settings.at

string | null

Extra model settings inferred from namespace

main.base_model_settings.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.base_model_settings.globally_disabled

boolean | null

default:

false

Whether the model is active for the org.

main.base_model_settings.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.base_model_settings.parent

object | null

main.base_model_settings.parent.added_by

string | null

required

The user (auth0_id) who created the model

main.base_model_settings.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

main.base_model_settings.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

main.base_model_settings.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

main.base_model_settings.parent.completion_token_cost

number

required

The cost of a completion token, in USD

main.base_model_settings.parent.context_window

integer

required

The max_tokens for this model

main.base_model_settings.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

main.base_model_settings.parent.description

string

required

A description of the model

main.base_model_settings.parent.id

string | null

required

The ID of this model

main.base_model_settings.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

main.base_model_settings.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

main.base_model_settings.parent.is_open_source

boolean

required

True if the model is open source

main.base_model_settings.parent.is_public

boolean

required

True if the model is publicly accessible to all

main.base_model_settings.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

main.base_model_settings.parent.is_rag

boolean

required

Whether it's rag-tuned or not

main.base_model_settings.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

main.base_model_settings.parent.model

string

required

The name of the model. Can belong to many providers

main.base_model_settings.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

main.base_model_settings.parent.namespace

string

required

The fully qualified (namespaced) model name

main.base_model_settings.parent.org_id

string | null

required

The org_id that has acccess to this model

main.base_model_settings.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

main.base_model_settings.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

main.base_model_settings.parent.supports_functions

boolean

required

True if the model supports function/tool call

main.base_model_settings.parent.supports_json

boolean

required

True if the model supports json-formatted responses

main.base_model_settings.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

main.base_model_settings.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

main.base_model_settings.parent.supports_stream

boolean

required

True if the model supports streaming responses

main.base_model_settings.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

main.base_model_settings.parent.until

string | null

required

The most recent data this model has been trained with

main.base_model_settings.parent.url

string

required

A URL to the model's page or more informatino

main.base_model_settings.parent.api_target

string

default:

Store the name of the model the API requires

main.base_model_settings.parent.at

string | null

Extra model settings inferred from namespace

main.base_model_settings.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.base_model_settings.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.base_model_settings.parent.parent

object | null

main.base_model_settings.parent.parent.added_by

string | null

required

The user (auth0_id) who created the model

main.base_model_settings.parent.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

main.base_model_settings.parent.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

main.base_model_settings.parent.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

main.base_model_settings.parent.parent.completion_token_cost

number

required

The cost of a completion token, in USD

main.base_model_settings.parent.parent.context_window

integer

required

The max_tokens for this model

main.base_model_settings.parent.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

main.base_model_settings.parent.parent.description

string

required

A description of the model

main.base_model_settings.parent.parent.id

string | null

required

The ID of this model

main.base_model_settings.parent.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

main.base_model_settings.parent.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

main.base_model_settings.parent.parent.is_open_source

boolean

required

True if the model is open source

main.base_model_settings.parent.parent.is_public

boolean

required

True if the model is publicly accessible to all

main.base_model_settings.parent.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

main.base_model_settings.parent.parent.is_rag

boolean

required

Whether it's rag-tuned or not

main.base_model_settings.parent.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

main.base_model_settings.parent.parent.model

string

required

The name of the model. Can belong to many providers

main.base_model_settings.parent.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

main.base_model_settings.parent.parent.namespace

string

required

The fully qualified (namespaced) model name

main.base_model_settings.parent.parent.org_id

string | null

required

The org_id that has acccess to this model

main.base_model_settings.parent.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

main.base_model_settings.parent.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

main.base_model_settings.parent.parent.supports_functions

boolean

required

True if the model supports function/tool call

main.base_model_settings.parent.parent.supports_json

boolean

required

True if the model supports json-formatted responses

main.base_model_settings.parent.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

main.base_model_settings.parent.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

main.base_model_settings.parent.parent.supports_stream

boolean

required

True if the model supports streaming responses

main.base_model_settings.parent.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

main.base_model_settings.parent.parent.until

string | null

required

The most recent data this model has been trained with

main.base_model_settings.parent.parent.url

string

required

A URL to the model's page or more informatino

main.base_model_settings.parent.parent.api_target

string

default:

Store the name of the model the API requires

main.base_model_settings.parent.parent.at

string | null

Extra model settings inferred from namespace

main.base_model_settings.parent.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.base_model_settings.parent.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.base_model_settings.parent.parent.parent

object | null

main.base_model_settings.parent.parent.parent_id

string | null

The ID of parent, in case it's not a base model

main.base_model_settings.parent.parent.prompt_id

string | null

The ID of prompt, used for this model

main.base_model_settings.parent.parent.provider

string | null

The provider for the model.

main.base_model_settings.parent.parent_id

string | null

The ID of parent, in case it's not a base model

main.base_model_settings.parent.prompt_id

string | null

The ID of prompt, used for this model

main.base_model_settings.parent.provider

string | null

The provider for the model.

main.base_model_settings.parent_id

string | null

The ID of parent, in case it's not a base model

main.base_model_settings.prompt_id

string | null

The ID of prompt, used for this model

main.base_model_settings.provider

string | null

The provider for the model.

main.custom_model_settings

object[]

required

main.custom_model_settings.context_window

integer

required

The max_tokens for this model

main.custom_model_settings.description

string

required

A description of the model

main.custom_model_settings.id

string | null

required

Used to uniquely target models when we enable/disable them

main.custom_model_settings.model

string

required

The name of the model. Can belong to many providers

main.custom_model_settings.namespace

string

required

The fully qualified (namespaced) model name

main.custom_model_settings.until

string | null

required

The most recent data this model has been trained with

main.custom_model_settings.url

string

required

A URL to the model's page or more informatino

main.custom_model_settings.active

boolean

default:

false

Whether the model is active for the app.

main.custom_model_settings.api_target

string

default:

Store the name of the model the API requires

main.custom_model_settings.at

string | null

Extra model settings inferred from namespace

main.custom_model_settings.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.custom_model_settings.globally_disabled

boolean | null

default:

false

Whether the model is active for the org.

main.custom_model_settings.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.custom_model_settings.parent

object | null

main.custom_model_settings.parent.added_by

string | null

required

The user (auth0_id) who created the model

main.custom_model_settings.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

main.custom_model_settings.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

main.custom_model_settings.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

main.custom_model_settings.parent.completion_token_cost

number

required

The cost of a completion token, in USD

main.custom_model_settings.parent.context_window

integer

required

The max_tokens for this model

main.custom_model_settings.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

main.custom_model_settings.parent.description

string

required

A description of the model

main.custom_model_settings.parent.id

string | null

required

The ID of this model

main.custom_model_settings.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

main.custom_model_settings.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

main.custom_model_settings.parent.is_open_source

boolean

required

True if the model is open source

main.custom_model_settings.parent.is_public

boolean

required

True if the model is publicly accessible to all

main.custom_model_settings.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

main.custom_model_settings.parent.is_rag

boolean

required

Whether it's rag-tuned or not

main.custom_model_settings.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

main.custom_model_settings.parent.model

string

required

The name of the model. Can belong to many providers

main.custom_model_settings.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

main.custom_model_settings.parent.namespace

string

required

The fully qualified (namespaced) model name

main.custom_model_settings.parent.org_id

string | null

required

The org_id that has acccess to this model

main.custom_model_settings.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

main.custom_model_settings.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

main.custom_model_settings.parent.supports_functions

boolean

required

True if the model supports function/tool call

main.custom_model_settings.parent.supports_json

boolean

required

True if the model supports json-formatted responses

main.custom_model_settings.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

main.custom_model_settings.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

main.custom_model_settings.parent.supports_stream

boolean

required

True if the model supports streaming responses

main.custom_model_settings.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

main.custom_model_settings.parent.until

string | null

required

The most recent data this model has been trained with

main.custom_model_settings.parent.url

string

required

A URL to the model's page or more informatino

main.custom_model_settings.parent.api_target

string

default:

Store the name of the model the API requires

main.custom_model_settings.parent.at

string | null

Extra model settings inferred from namespace

main.custom_model_settings.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.custom_model_settings.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.custom_model_settings.parent.parent

object | null

main.custom_model_settings.parent.parent.added_by

string | null

required

The user (auth0_id) who created the model

main.custom_model_settings.parent.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

main.custom_model_settings.parent.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

main.custom_model_settings.parent.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

main.custom_model_settings.parent.parent.completion_token_cost

number

required

The cost of a completion token, in USD

main.custom_model_settings.parent.parent.context_window

integer

required

The max_tokens for this model

main.custom_model_settings.parent.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

main.custom_model_settings.parent.parent.description

string

required

A description of the model

main.custom_model_settings.parent.parent.id

string | null

required

The ID of this model

main.custom_model_settings.parent.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

main.custom_model_settings.parent.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

main.custom_model_settings.parent.parent.is_open_source

boolean

required

True if the model is open source

main.custom_model_settings.parent.parent.is_public

boolean

required

True if the model is publicly accessible to all

main.custom_model_settings.parent.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

main.custom_model_settings.parent.parent.is_rag

boolean

required

Whether it's rag-tuned or not

main.custom_model_settings.parent.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

main.custom_model_settings.parent.parent.model

string

required

The name of the model. Can belong to many providers

main.custom_model_settings.parent.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

main.custom_model_settings.parent.parent.namespace

string

required

The fully qualified (namespaced) model name

main.custom_model_settings.parent.parent.org_id

string | null

required

The org_id that has acccess to this model

main.custom_model_settings.parent.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

main.custom_model_settings.parent.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

main.custom_model_settings.parent.parent.supports_functions

boolean

required

True if the model supports function/tool call

main.custom_model_settings.parent.parent.supports_json

boolean

required

True if the model supports json-formatted responses

main.custom_model_settings.parent.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

main.custom_model_settings.parent.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

main.custom_model_settings.parent.parent.supports_stream

boolean

required

True if the model supports streaming responses

main.custom_model_settings.parent.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

main.custom_model_settings.parent.parent.until

string | null

required

The most recent data this model has been trained with

main.custom_model_settings.parent.parent.url

string

required

A URL to the model's page or more informatino

main.custom_model_settings.parent.parent.api_target

string

default:

Store the name of the model the API requires

main.custom_model_settings.parent.parent.at

string | null

Extra model settings inferred from namespace

main.custom_model_settings.parent.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.custom_model_settings.parent.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.custom_model_settings.parent.parent.parent

object | null

main.custom_model_settings.parent.parent.parent_id

string | null

The ID of parent, in case it's not a base model

main.custom_model_settings.parent.parent.prompt_id

string | null

The ID of prompt, used for this model

main.custom_model_settings.parent.parent.provider

string | null

The provider for the model.

main.custom_model_settings.parent.parent_id

string | null

The ID of parent, in case it's not a base model

main.custom_model_settings.parent.prompt_id

string | null

The ID of prompt, used for this model

main.custom_model_settings.parent.provider

string | null

The provider for the model.

main.custom_model_settings.parent_id

string | null

The ID of parent, in case it's not a base model

main.custom_model_settings.prompt_id

string | null

The ID of prompt, used for this model

main.custom_model_settings.provider

string | null

The provider for the model.

main.failover_models

object[]

required

main.failover_models.added_by

string | null

required

The user (auth0_id) who created the model

main.failover_models.added_on

string | null

required

When the model was added. Auto-populated in DB

main.failover_models.app_id

string | null

required

The app_id that has access to this model (if only one)

main.failover_models.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

main.failover_models.completion_token_cost

number

required

The cost of a completion token, in USD

main.failover_models.context_window

integer

required

The max_tokens for this model

main.failover_models.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

main.failover_models.description

string

required

A description of the model

main.failover_models.id

string | null

required

The ID of this model

main.failover_models.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

main.failover_models.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

main.failover_models.is_open_source

boolean

required

True if the model is open source

main.failover_models.is_public

boolean

required

True if the model is publicly accessible to all

main.failover_models.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

main.failover_models.is_rag

boolean

required

Whether it's rag-tuned or not

main.failover_models.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

main.failover_models.model

string

required

The name of the model. Can belong to many providers

main.failover_models.modified_on

string | null

required

When the model was updated. Auto-populated in DB

main.failover_models.namespace

string

required

The fully qualified (namespaced) model name

main.failover_models.org_id

string | null

required

The org_id that has acccess to this model

main.failover_models.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

main.failover_models.prompt_token_cost

number

required

The cost of a prompt token, in USD

main.failover_models.supports_functions

boolean

required

True if the model supports function/tool call

main.failover_models.supports_json

boolean

required

True if the model supports json-formatted responses

main.failover_models.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

main.failover_models.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

main.failover_models.supports_stream

boolean

required

True if the model supports streaming responses

main.failover_models.supports_vision

boolean

required

True if the model supports image recognition (vision)

main.failover_models.until

string | null

required

The most recent data this model has been trained with

main.failover_models.url

string

required

A URL to the model's page or more informatino

main.failover_models.api_target

string

default:

Store the name of the model the API requires

main.failover_models.at

string | null

Extra model settings inferred from namespace

main.failover_models.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.failover_models.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.failover_models.parent

object | null

main.failover_models.parent.added_by

string | null

required

The user (auth0_id) who created the model

main.failover_models.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

main.failover_models.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

main.failover_models.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

main.failover_models.parent.completion_token_cost

number

required

The cost of a completion token, in USD

main.failover_models.parent.context_window

integer

required

The max_tokens for this model

main.failover_models.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

main.failover_models.parent.description

string

required

A description of the model

main.failover_models.parent.id

string | null

required

The ID of this model

main.failover_models.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

main.failover_models.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

main.failover_models.parent.is_open_source

boolean

required

True if the model is open source

main.failover_models.parent.is_public

boolean

required

True if the model is publicly accessible to all

main.failover_models.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

main.failover_models.parent.is_rag

boolean

required

Whether it's rag-tuned or not

main.failover_models.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

main.failover_models.parent.model

string

required

The name of the model. Can belong to many providers

main.failover_models.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

main.failover_models.parent.namespace

string

required

The fully qualified (namespaced) model name

main.failover_models.parent.org_id

string | null

required

The org_id that has acccess to this model

main.failover_models.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

main.failover_models.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

main.failover_models.parent.supports_functions

boolean

required

True if the model supports function/tool call

main.failover_models.parent.supports_json

boolean

required

True if the model supports json-formatted responses

main.failover_models.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

main.failover_models.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

main.failover_models.parent.supports_stream

boolean

required

True if the model supports streaming responses

main.failover_models.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

main.failover_models.parent.until

string | null

required

The most recent data this model has been trained with

main.failover_models.parent.url

string

required

A URL to the model's page or more informatino

main.failover_models.parent.api_target

string

default:

Store the name of the model the API requires

main.failover_models.parent.at

string | null

Extra model settings inferred from namespace

main.failover_models.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.failover_models.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.failover_models.parent.parent

object | null

main.failover_models.parent.parent.added_by

string | null

required

The user (auth0_id) who created the model

main.failover_models.parent.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

main.failover_models.parent.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

main.failover_models.parent.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

main.failover_models.parent.parent.completion_token_cost

number

required

The cost of a completion token, in USD

main.failover_models.parent.parent.context_window

integer

required

The max_tokens for this model

main.failover_models.parent.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

main.failover_models.parent.parent.description

string

required

A description of the model

main.failover_models.parent.parent.id

string | null

required

The ID of this model

main.failover_models.parent.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

main.failover_models.parent.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

main.failover_models.parent.parent.is_open_source

boolean

required

True if the model is open source

main.failover_models.parent.parent.is_public

boolean

required

True if the model is publicly accessible to all

main.failover_models.parent.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

main.failover_models.parent.parent.is_rag

boolean

required

Whether it's rag-tuned or not

main.failover_models.parent.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

main.failover_models.parent.parent.model

string

required

The name of the model. Can belong to many providers

main.failover_models.parent.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

main.failover_models.parent.parent.namespace

string

required

The fully qualified (namespaced) model name

main.failover_models.parent.parent.org_id

string | null

required

The org_id that has acccess to this model

main.failover_models.parent.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

main.failover_models.parent.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

main.failover_models.parent.parent.supports_functions

boolean

required

True if the model supports function/tool call

main.failover_models.parent.parent.supports_json

boolean

required

True if the model supports json-formatted responses

main.failover_models.parent.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

main.failover_models.parent.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

main.failover_models.parent.parent.supports_stream

boolean

required

True if the model supports streaming responses

main.failover_models.parent.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

main.failover_models.parent.parent.until

string | null

required

The most recent data this model has been trained with

main.failover_models.parent.parent.url

string

required

A URL to the model's page or more informatino

main.failover_models.parent.parent.api_target

string

default:

Store the name of the model the API requires

main.failover_models.parent.parent.at

string | null

Extra model settings inferred from namespace

main.failover_models.parent.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

main.failover_models.parent.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

main.failover_models.parent.parent.parent

object | null

main.failover_models.parent.parent.parent_id

string | null

The ID of parent, in case it's not a base model

main.failover_models.parent.parent.prompt_id

string | null

The ID of prompt, used for this model

main.failover_models.parent.parent.provider

string | null

The provider for the model.

main.failover_models.parent.parent_id

string | null

The ID of parent, in case it's not a base model

main.failover_models.parent.prompt_id

string | null

The ID of prompt, used for this model

main.failover_models.parent.provider

string | null

The provider for the model.

main.failover_models.parent_id

string | null

The ID of parent, in case it's not a base model

main.failover_models.prompt_id

string | null

The ID of prompt, used for this model

main.failover_models.provider

string | null

The provider for the model.

main.current_user_permissions

string[] | null

sandbox

object | null

required

sandbox.active_models

integer

required

sandbox.app

object

required

sandbox.app.auth0_id

string

required

The Auth0ID of the creator of the app

sandbox.app.benchmark_model

object

required

Compare the results of LLMs against this model (for speed, quality, cost...)

sandbox.app.benchmark_model.added_by

string | null

required

The user (auth0_id) who created the model

sandbox.app.benchmark_model.added_on

string | null

required

When the model was added. Auto-populated in DB

sandbox.app.benchmark_model.app_id

string | null

required

The app_id that has access to this model (if only one)

sandbox.app.benchmark_model.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

sandbox.app.benchmark_model.completion_token_cost

number

required

The cost of a completion token, in USD

sandbox.app.benchmark_model.context_window

integer

required

The max_tokens for this model

sandbox.app.benchmark_model.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

sandbox.app.benchmark_model.description

string

required

A description of the model

sandbox.app.benchmark_model.id

string | null

required

The ID of this model

sandbox.app.benchmark_model.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

sandbox.app.benchmark_model.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

sandbox.app.benchmark_model.is_open_source

boolean

required

True if the model is open source

sandbox.app.benchmark_model.is_public

boolean

required

True if the model is publicly accessible to all

sandbox.app.benchmark_model.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

sandbox.app.benchmark_model.is_rag

boolean

required

Whether it's rag-tuned or not

sandbox.app.benchmark_model.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

sandbox.app.benchmark_model.model

string

required

The name of the model. Can belong to many providers

sandbox.app.benchmark_model.modified_on

string | null

required

When the model was updated. Auto-populated in DB

sandbox.app.benchmark_model.namespace

string

required

The fully qualified (namespaced) model name

sandbox.app.benchmark_model.org_id

string | null

required

The org_id that has acccess to this model

sandbox.app.benchmark_model.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

sandbox.app.benchmark_model.prompt_token_cost

number

required

The cost of a prompt token, in USD

sandbox.app.benchmark_model.supports_functions

boolean

required

True if the model supports function/tool call

sandbox.app.benchmark_model.supports_json

boolean

required

True if the model supports json-formatted responses

sandbox.app.benchmark_model.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

sandbox.app.benchmark_model.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

sandbox.app.benchmark_model.supports_stream

boolean

required

True if the model supports streaming responses

sandbox.app.benchmark_model.supports_vision

boolean

required

True if the model supports image recognition (vision)

sandbox.app.benchmark_model.until

string | null

required

The most recent data this model has been trained with

sandbox.app.benchmark_model.url

string

required

A URL to the model's page or more informatino

sandbox.app.benchmark_model.api_target

string

default:

Store the name of the model the API requires

sandbox.app.benchmark_model.at

string | null

Extra model settings inferred from namespace

sandbox.app.benchmark_model.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.app.benchmark_model.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.app.benchmark_model.parent

object | null

sandbox.app.benchmark_model.parent.added_by

string | null

required

The user (auth0_id) who created the model

sandbox.app.benchmark_model.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

sandbox.app.benchmark_model.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

sandbox.app.benchmark_model.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

sandbox.app.benchmark_model.parent.completion_token_cost

number

required

The cost of a completion token, in USD

sandbox.app.benchmark_model.parent.context_window

integer

required

The max_tokens for this model

sandbox.app.benchmark_model.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

sandbox.app.benchmark_model.parent.description

string

required

A description of the model

sandbox.app.benchmark_model.parent.id

string | null

required

The ID of this model

sandbox.app.benchmark_model.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

sandbox.app.benchmark_model.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

sandbox.app.benchmark_model.parent.is_open_source

boolean

required

True if the model is open source

sandbox.app.benchmark_model.parent.is_public

boolean

required

True if the model is publicly accessible to all

sandbox.app.benchmark_model.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

sandbox.app.benchmark_model.parent.is_rag

boolean

required

Whether it's rag-tuned or not

sandbox.app.benchmark_model.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

sandbox.app.benchmark_model.parent.model

string

required

The name of the model. Can belong to many providers

sandbox.app.benchmark_model.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

sandbox.app.benchmark_model.parent.namespace

string

required

The fully qualified (namespaced) model name

sandbox.app.benchmark_model.parent.org_id

string | null

required

The org_id that has acccess to this model

sandbox.app.benchmark_model.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

sandbox.app.benchmark_model.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

sandbox.app.benchmark_model.parent.supports_functions

boolean

required

True if the model supports function/tool call

sandbox.app.benchmark_model.parent.supports_json

boolean

required

True if the model supports json-formatted responses

sandbox.app.benchmark_model.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

sandbox.app.benchmark_model.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

sandbox.app.benchmark_model.parent.supports_stream

boolean

required

True if the model supports streaming responses

sandbox.app.benchmark_model.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

sandbox.app.benchmark_model.parent.until

string | null

required

The most recent data this model has been trained with

sandbox.app.benchmark_model.parent.url

string

required

A URL to the model's page or more informatino

sandbox.app.benchmark_model.parent.api_target

string

default:

Store the name of the model the API requires

sandbox.app.benchmark_model.parent.at

string | null

Extra model settings inferred from namespace

sandbox.app.benchmark_model.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.app.benchmark_model.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.app.benchmark_model.parent.parent

object | null

sandbox.app.benchmark_model.parent.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.app.benchmark_model.parent.prompt_id

string | null

The ID of prompt, used for this model

sandbox.app.benchmark_model.parent.provider

string | null

The provider for the model.

sandbox.app.benchmark_model.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.app.benchmark_model.prompt_id

string | null

The ID of prompt, used for this model

sandbox.app.benchmark_model.provider

string | null

The provider for the model.

sandbox.app.benchmark_model_id

string

required

Compare the results of LLMs against this model (for speed, quality, cost...)

sandbox.app.has_failover_chain

boolean

required

A Failover chain, when enabled, skips the SMART router and instead calls the failover chain in order.

sandbox.app.hashed_key

string

required

The hashed value of the API Key which we store in our database

sandbox.app.id

string

required

sandbox.app.is_active

boolean

required

True if the app is active (not soft-deleted, etc.), false otherwise

sandbox.app.key_end

string

required

The last characters of the API key

sandbox.app.modified_by

string

required

The Auth0ID of the last person to modify the table

sandbox.app.name

string

required

A name for this app

sandbox.app.prompt

object | null

required

The Prompt associated to the app, if any

sandbox.app.added_on

string

sandbox.app.app_logo

string | null

A logo for this app

sandbox.app.description

string | null

A description for this app

sandbox.app.modified_on

string

sandbox.app.org_id

string

The Org to which this App belongs

sandbox.app.policies

object

sandbox.app.policies.max_cost

number

default:

0.02

The maximum cost allowed for a request. Only works with compounded requests that require multiple LLM calls. If the value is reached, it will exit with an exception.

Required range: x > 0.0001

sandbox.app.policies.max_same_model_retries

integer

default:

If an LLM call fails, how many times should Pulze retry the call to the same LLM? There will be a maximum of N+1 calls (original + N retries)

Required range: 0 < x < 3

sandbox.app.policies.max_switch_model_retries

integer

default:

If an LLM call fails, how many other models should Pulze try, chosen by quality descending? It will be a maximum of N+1 models (original + N other models)

Required range: 0 < x < 5

sandbox.app.policies.optimize_internal_requests

enum<integer>

default:

Available options:

0,

1

sandbox.app.policies.privacy_level

enum<integer>

default:

    The level of privacy for a given request
    0 = (UNSUPPORTED -- public logs)
    1 = Log request, response and all of its metadata (Normal mode)
    2 = Do not log neither the request prompt nor the response text. Logs are still visible, and all of the request metadata accessible. Retrievable as a log. (TBD)
    3 = Do not log at all. Internally, a minimal representation may be stored for billing: model name, tokens used, which app it belongs to, and timestamp. Not retrievable as a log. (TBD)

Available options:

1,

2,

3

sandbox.app.policies.prompt_id

string | null

Prompt ID that we will use for requests

sandbox.app.prompt_id

string | null

If any, the ID of the prompt associated with this app

sandbox.app.rate_limit

integer

default:

200

sandbox.app.sandbox_of

string | null

The app might be a (testing) sandbox of a different App.

sandbox.app.settings

object

sandbox.app.use_parent_custom_data

boolean | null

If true, the app will use the custom data of the parent app

sandbox.app.weights

object

sandbox.base_model_settings

object[]

required

sandbox.base_model_settings.context_window

integer

required

The max_tokens for this model

sandbox.base_model_settings.description

string

required

A description of the model

sandbox.base_model_settings.id

string | null

required

Used to uniquely target models when we enable/disable them

sandbox.base_model_settings.model

string

required

The name of the model. Can belong to many providers

sandbox.base_model_settings.namespace

string

required

The fully qualified (namespaced) model name

sandbox.base_model_settings.until

string | null

required

The most recent data this model has been trained with

sandbox.base_model_settings.url

string

required

A URL to the model's page or more informatino

sandbox.base_model_settings.active

boolean

default:

false

Whether the model is active for the app.

sandbox.base_model_settings.api_target

string

default:

Store the name of the model the API requires

sandbox.base_model_settings.at

string | null

Extra model settings inferred from namespace

sandbox.base_model_settings.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.base_model_settings.globally_disabled

boolean | null

default:

false

Whether the model is active for the org.

sandbox.base_model_settings.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.base_model_settings.parent

object | null

sandbox.base_model_settings.parent.added_by

string | null

required

The user (auth0_id) who created the model

sandbox.base_model_settings.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

sandbox.base_model_settings.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

sandbox.base_model_settings.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

sandbox.base_model_settings.parent.completion_token_cost

number

required

The cost of a completion token, in USD

sandbox.base_model_settings.parent.context_window

integer

required

The max_tokens for this model

sandbox.base_model_settings.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

sandbox.base_model_settings.parent.description

string

required

A description of the model

sandbox.base_model_settings.parent.id

string | null

required

The ID of this model

sandbox.base_model_settings.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

sandbox.base_model_settings.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

sandbox.base_model_settings.parent.is_open_source

boolean

required

True if the model is open source

sandbox.base_model_settings.parent.is_public

boolean

required

True if the model is publicly accessible to all

sandbox.base_model_settings.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

sandbox.base_model_settings.parent.is_rag

boolean

required

Whether it's rag-tuned or not

sandbox.base_model_settings.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

sandbox.base_model_settings.parent.model

string

required

The name of the model. Can belong to many providers

sandbox.base_model_settings.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

sandbox.base_model_settings.parent.namespace

string

required

The fully qualified (namespaced) model name

sandbox.base_model_settings.parent.org_id

string | null

required

The org_id that has acccess to this model

sandbox.base_model_settings.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

sandbox.base_model_settings.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

sandbox.base_model_settings.parent.supports_functions

boolean

required

True if the model supports function/tool call

sandbox.base_model_settings.parent.supports_json

boolean

required

True if the model supports json-formatted responses

sandbox.base_model_settings.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

sandbox.base_model_settings.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

sandbox.base_model_settings.parent.supports_stream

boolean

required

True if the model supports streaming responses

sandbox.base_model_settings.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

sandbox.base_model_settings.parent.until

string | null

required

The most recent data this model has been trained with

sandbox.base_model_settings.parent.url

string

required

A URL to the model's page or more informatino

sandbox.base_model_settings.parent.api_target

string

default:

Store the name of the model the API requires

sandbox.base_model_settings.parent.at

string | null

Extra model settings inferred from namespace

sandbox.base_model_settings.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.base_model_settings.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.base_model_settings.parent.parent

object | null

sandbox.base_model_settings.parent.parent.added_by

string | null

required

The user (auth0_id) who created the model

sandbox.base_model_settings.parent.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

sandbox.base_model_settings.parent.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

sandbox.base_model_settings.parent.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

sandbox.base_model_settings.parent.parent.completion_token_cost

number

required

The cost of a completion token, in USD

sandbox.base_model_settings.parent.parent.context_window

integer

required

The max_tokens for this model

sandbox.base_model_settings.parent.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

sandbox.base_model_settings.parent.parent.description

string

required

A description of the model

sandbox.base_model_settings.parent.parent.id

string | null

required

The ID of this model

sandbox.base_model_settings.parent.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

sandbox.base_model_settings.parent.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

sandbox.base_model_settings.parent.parent.is_open_source

boolean

required

True if the model is open source

sandbox.base_model_settings.parent.parent.is_public

boolean

required

True if the model is publicly accessible to all

sandbox.base_model_settings.parent.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

sandbox.base_model_settings.parent.parent.is_rag

boolean

required

Whether it's rag-tuned or not

sandbox.base_model_settings.parent.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

sandbox.base_model_settings.parent.parent.model

string

required

The name of the model. Can belong to many providers

sandbox.base_model_settings.parent.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

sandbox.base_model_settings.parent.parent.namespace

string

required

The fully qualified (namespaced) model name

sandbox.base_model_settings.parent.parent.org_id

string | null

required

The org_id that has acccess to this model

sandbox.base_model_settings.parent.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

sandbox.base_model_settings.parent.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

sandbox.base_model_settings.parent.parent.supports_functions

boolean

required

True if the model supports function/tool call

sandbox.base_model_settings.parent.parent.supports_json

boolean

required

True if the model supports json-formatted responses

sandbox.base_model_settings.parent.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

sandbox.base_model_settings.parent.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

sandbox.base_model_settings.parent.parent.supports_stream

boolean

required

True if the model supports streaming responses

sandbox.base_model_settings.parent.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

sandbox.base_model_settings.parent.parent.until

string | null

required

The most recent data this model has been trained with

sandbox.base_model_settings.parent.parent.url

string

required

A URL to the model's page or more informatino

sandbox.base_model_settings.parent.parent.api_target

string

default:

Store the name of the model the API requires

sandbox.base_model_settings.parent.parent.at

string | null

Extra model settings inferred from namespace

sandbox.base_model_settings.parent.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.base_model_settings.parent.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.base_model_settings.parent.parent.parent

object | null

sandbox.base_model_settings.parent.parent.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.base_model_settings.parent.parent.prompt_id

string | null

The ID of prompt, used for this model

sandbox.base_model_settings.parent.parent.provider

string | null

The provider for the model.

sandbox.base_model_settings.parent.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.base_model_settings.parent.prompt_id

string | null

The ID of prompt, used for this model

sandbox.base_model_settings.parent.provider

string | null

The provider for the model.

sandbox.base_model_settings.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.base_model_settings.prompt_id

string | null

The ID of prompt, used for this model

sandbox.base_model_settings.provider

string | null

The provider for the model.

sandbox.custom_model_settings

object[]

required

sandbox.custom_model_settings.context_window

integer

required

The max_tokens for this model

sandbox.custom_model_settings.description

string

required

A description of the model

sandbox.custom_model_settings.id

string | null

required

Used to uniquely target models when we enable/disable them

sandbox.custom_model_settings.model

string

required

The name of the model. Can belong to many providers

sandbox.custom_model_settings.namespace

string

required

The fully qualified (namespaced) model name

sandbox.custom_model_settings.until

string | null

required

The most recent data this model has been trained with

sandbox.custom_model_settings.url

string

required

A URL to the model's page or more informatino

sandbox.custom_model_settings.active

boolean

default:

false

Whether the model is active for the app.

sandbox.custom_model_settings.api_target

string

default:

Store the name of the model the API requires

sandbox.custom_model_settings.at

string | null

Extra model settings inferred from namespace

sandbox.custom_model_settings.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.custom_model_settings.globally_disabled

boolean | null

default:

false

Whether the model is active for the org.

sandbox.custom_model_settings.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.custom_model_settings.parent

object | null

sandbox.custom_model_settings.parent.added_by

string | null

required

The user (auth0_id) who created the model

sandbox.custom_model_settings.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

sandbox.custom_model_settings.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

sandbox.custom_model_settings.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

sandbox.custom_model_settings.parent.completion_token_cost

number

required

The cost of a completion token, in USD

sandbox.custom_model_settings.parent.context_window

integer

required

The max_tokens for this model

sandbox.custom_model_settings.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

sandbox.custom_model_settings.parent.description

string

required

A description of the model

sandbox.custom_model_settings.parent.id

string | null

required

The ID of this model

sandbox.custom_model_settings.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

sandbox.custom_model_settings.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

sandbox.custom_model_settings.parent.is_open_source

boolean

required

True if the model is open source

sandbox.custom_model_settings.parent.is_public

boolean

required

True if the model is publicly accessible to all

sandbox.custom_model_settings.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

sandbox.custom_model_settings.parent.is_rag

boolean

required

Whether it's rag-tuned or not

sandbox.custom_model_settings.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

sandbox.custom_model_settings.parent.model

string

required

The name of the model. Can belong to many providers

sandbox.custom_model_settings.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

sandbox.custom_model_settings.parent.namespace

string

required

The fully qualified (namespaced) model name

sandbox.custom_model_settings.parent.org_id

string | null

required

The org_id that has acccess to this model

sandbox.custom_model_settings.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

sandbox.custom_model_settings.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

sandbox.custom_model_settings.parent.supports_functions

boolean

required

True if the model supports function/tool call

sandbox.custom_model_settings.parent.supports_json

boolean

required

True if the model supports json-formatted responses

sandbox.custom_model_settings.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

sandbox.custom_model_settings.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

sandbox.custom_model_settings.parent.supports_stream

boolean

required

True if the model supports streaming responses

sandbox.custom_model_settings.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

sandbox.custom_model_settings.parent.until

string | null

required

The most recent data this model has been trained with

sandbox.custom_model_settings.parent.url

string

required

A URL to the model's page or more informatino

sandbox.custom_model_settings.parent.api_target

string

default:

Store the name of the model the API requires

sandbox.custom_model_settings.parent.at

string | null

Extra model settings inferred from namespace

sandbox.custom_model_settings.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.custom_model_settings.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.custom_model_settings.parent.parent

object | null

sandbox.custom_model_settings.parent.parent.added_by

string | null

required

The user (auth0_id) who created the model

sandbox.custom_model_settings.parent.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

sandbox.custom_model_settings.parent.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

sandbox.custom_model_settings.parent.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

sandbox.custom_model_settings.parent.parent.completion_token_cost

number

required

The cost of a completion token, in USD

sandbox.custom_model_settings.parent.parent.context_window

integer

required

The max_tokens for this model

sandbox.custom_model_settings.parent.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

sandbox.custom_model_settings.parent.parent.description

string

required

A description of the model

sandbox.custom_model_settings.parent.parent.id

string | null

required

The ID of this model

sandbox.custom_model_settings.parent.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

sandbox.custom_model_settings.parent.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

sandbox.custom_model_settings.parent.parent.is_open_source

boolean

required

True if the model is open source

sandbox.custom_model_settings.parent.parent.is_public

boolean

required

True if the model is publicly accessible to all

sandbox.custom_model_settings.parent.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

sandbox.custom_model_settings.parent.parent.is_rag

boolean

required

Whether it's rag-tuned or not

sandbox.custom_model_settings.parent.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

sandbox.custom_model_settings.parent.parent.model

string

required

The name of the model. Can belong to many providers

sandbox.custom_model_settings.parent.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

sandbox.custom_model_settings.parent.parent.namespace

string

required

The fully qualified (namespaced) model name

sandbox.custom_model_settings.parent.parent.org_id

string | null

required

The org_id that has acccess to this model

sandbox.custom_model_settings.parent.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

sandbox.custom_model_settings.parent.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

sandbox.custom_model_settings.parent.parent.supports_functions

boolean

required

True if the model supports function/tool call

sandbox.custom_model_settings.parent.parent.supports_json

boolean

required

True if the model supports json-formatted responses

sandbox.custom_model_settings.parent.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

sandbox.custom_model_settings.parent.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

sandbox.custom_model_settings.parent.parent.supports_stream

boolean

required

True if the model supports streaming responses

sandbox.custom_model_settings.parent.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

sandbox.custom_model_settings.parent.parent.until

string | null

required

The most recent data this model has been trained with

sandbox.custom_model_settings.parent.parent.url

string

required

A URL to the model's page or more informatino

sandbox.custom_model_settings.parent.parent.api_target

string

default:

Store the name of the model the API requires

sandbox.custom_model_settings.parent.parent.at

string | null

Extra model settings inferred from namespace

sandbox.custom_model_settings.parent.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.custom_model_settings.parent.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.custom_model_settings.parent.parent.parent

object | null

sandbox.custom_model_settings.parent.parent.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.custom_model_settings.parent.parent.prompt_id

string | null

The ID of prompt, used for this model

sandbox.custom_model_settings.parent.parent.provider

string | null

The provider for the model.

sandbox.custom_model_settings.parent.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.custom_model_settings.parent.prompt_id

string | null

The ID of prompt, used for this model

sandbox.custom_model_settings.parent.provider

string | null

The provider for the model.

sandbox.custom_model_settings.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.custom_model_settings.prompt_id

string | null

The ID of prompt, used for this model

sandbox.custom_model_settings.provider

string | null

The provider for the model.

sandbox.failover_models

object[]

required

sandbox.failover_models.added_by

string | null

required

The user (auth0_id) who created the model

sandbox.failover_models.added_on

string | null

required

When the model was added. Auto-populated in DB

sandbox.failover_models.app_id

string | null

required

The app_id that has access to this model (if only one)

sandbox.failover_models.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

sandbox.failover_models.completion_token_cost

number

required

The cost of a completion token, in USD

sandbox.failover_models.context_window

integer

required

The max_tokens for this model

sandbox.failover_models.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

sandbox.failover_models.description

string

required

A description of the model

sandbox.failover_models.id

string | null

required

The ID of this model

sandbox.failover_models.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

sandbox.failover_models.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

sandbox.failover_models.is_open_source

boolean

required

True if the model is open source

sandbox.failover_models.is_public

boolean

required

True if the model is publicly accessible to all

sandbox.failover_models.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

sandbox.failover_models.is_rag

boolean

required

Whether it's rag-tuned or not

sandbox.failover_models.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

sandbox.failover_models.model

string

required

The name of the model. Can belong to many providers

sandbox.failover_models.modified_on

string | null

required

When the model was updated. Auto-populated in DB

sandbox.failover_models.namespace

string

required

The fully qualified (namespaced) model name

sandbox.failover_models.org_id

string | null

required

The org_id that has acccess to this model

sandbox.failover_models.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

sandbox.failover_models.prompt_token_cost

number

required

The cost of a prompt token, in USD

sandbox.failover_models.supports_functions

boolean

required

True if the model supports function/tool call

sandbox.failover_models.supports_json

boolean

required

True if the model supports json-formatted responses

sandbox.failover_models.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

sandbox.failover_models.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

sandbox.failover_models.supports_stream

boolean

required

True if the model supports streaming responses

sandbox.failover_models.supports_vision

boolean

required

True if the model supports image recognition (vision)

sandbox.failover_models.until

string | null

required

The most recent data this model has been trained with

sandbox.failover_models.url

string

required

A URL to the model's page or more informatino

sandbox.failover_models.api_target

string

default:

Store the name of the model the API requires

sandbox.failover_models.at

string | null

Extra model settings inferred from namespace

sandbox.failover_models.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.failover_models.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.failover_models.parent

object | null

sandbox.failover_models.parent.added_by

string | null

required

The user (auth0_id) who created the model

sandbox.failover_models.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

sandbox.failover_models.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

sandbox.failover_models.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

sandbox.failover_models.parent.completion_token_cost

number

required

The cost of a completion token, in USD

sandbox.failover_models.parent.context_window

integer

required

The max_tokens for this model

sandbox.failover_models.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

sandbox.failover_models.parent.description

string

required

A description of the model

sandbox.failover_models.parent.id

string | null

required

The ID of this model

sandbox.failover_models.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

sandbox.failover_models.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

sandbox.failover_models.parent.is_open_source

boolean

required

True if the model is open source

sandbox.failover_models.parent.is_public

boolean

required

True if the model is publicly accessible to all

sandbox.failover_models.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

sandbox.failover_models.parent.is_rag

boolean

required

Whether it's rag-tuned or not

sandbox.failover_models.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

sandbox.failover_models.parent.model

string

required

The name of the model. Can belong to many providers

sandbox.failover_models.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

sandbox.failover_models.parent.namespace

string

required

The fully qualified (namespaced) model name

sandbox.failover_models.parent.org_id

string | null

required

The org_id that has acccess to this model

sandbox.failover_models.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

sandbox.failover_models.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

sandbox.failover_models.parent.supports_functions

boolean

required

True if the model supports function/tool call

sandbox.failover_models.parent.supports_json

boolean

required

True if the model supports json-formatted responses

sandbox.failover_models.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

sandbox.failover_models.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

sandbox.failover_models.parent.supports_stream

boolean

required

True if the model supports streaming responses

sandbox.failover_models.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

sandbox.failover_models.parent.until

string | null

required

The most recent data this model has been trained with

sandbox.failover_models.parent.url

string

required

A URL to the model's page or more informatino

sandbox.failover_models.parent.api_target

string

default:

Store the name of the model the API requires

sandbox.failover_models.parent.at

string | null

Extra model settings inferred from namespace

sandbox.failover_models.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.failover_models.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.failover_models.parent.parent

object | null

sandbox.failover_models.parent.parent.added_by

string | null

required

The user (auth0_id) who created the model

sandbox.failover_models.parent.parent.added_on

string | null

required

When the model was added. Auto-populated in DB

sandbox.failover_models.parent.parent.app_id

string | null

required

The app_id that has access to this model (if only one)

sandbox.failover_models.parent.parent.base_cost

number

required

A (usually 0) cost added on top of a request. Some models charge per request, not only per token

sandbox.failover_models.parent.parent.completion_token_cost

number

required

The cost of a completion token, in USD

sandbox.failover_models.parent.parent.context_window

integer

required

The max_tokens for this model

sandbox.failover_models.parent.parent.default_active

boolean

required

This determines if the model will be available + pre-selected when users create new apps.

sandbox.failover_models.parent.parent.description

string

required

A description of the model

sandbox.failover_models.parent.parent.id

string | null

required

The ID of this model

sandbox.failover_models.parent.parent.is_chat

boolean

required

True if the model is of type Chat Completions, False if it's a Text Completion model.

sandbox.failover_models.parent.parent.is_ft

boolean

required

Whether it's fine-tuned or not

True if the model complies with GDPR

sandbox.failover_models.parent.parent.is_open_source

boolean

required

True if the model is open source

sandbox.failover_models.parent.parent.is_public

boolean

required

True if the model is publicly accessible to all

sandbox.failover_models.parent.parent.is_pulze_owner

boolean

required

Model has been created and shared by Pulze

sandbox.failover_models.parent.parent.is_rag

boolean

required

Whether it's rag-tuned or not

sandbox.failover_models.parent.parent.is_test_model

boolean

required

Test models are only used for testing and do not perform any LLM requests

sandbox.failover_models.parent.parent.model

string

required

The name of the model. Can belong to many providers

sandbox.failover_models.parent.parent.modified_on

string | null

required

When the model was updated. Auto-populated in DB

sandbox.failover_models.parent.parent.namespace

string

required

The fully qualified (namespaced) model name

sandbox.failover_models.parent.parent.org_id

string | null

required

The org_id that has acccess to this model

sandbox.failover_models.parent.parent.price_unit

enum<string>

required

The unit of billing for this model

Available options:

tokens,

characters

sandbox.failover_models.parent.parent.prompt_token_cost

number

required

The cost of a prompt token, in USD

sandbox.failover_models.parent.parent.supports_functions

boolean

required

True if the model supports function/tool call

sandbox.failover_models.parent.parent.supports_json

boolean

required

True if the model supports json-formatted responses

sandbox.failover_models.parent.parent.supports_n

boolean

required

True if the model supports n and best_of -- i.e, multiple responses

sandbox.failover_models.parent.parent.supports_penalties

boolean

required

True if the model supports frequency_penalty and presence_penalty

sandbox.failover_models.parent.parent.supports_stream

boolean

required

True if the model supports streaming responses

sandbox.failover_models.parent.parent.supports_vision

boolean

required

True if the model supports image recognition (vision)

sandbox.failover_models.parent.parent.until

string | null

required

The most recent data this model has been trained with

sandbox.failover_models.parent.parent.url

string

required

A URL to the model's page or more informatino

sandbox.failover_models.parent.parent.api_target

string

default:

Store the name of the model the API requires

sandbox.failover_models.parent.parent.at

string | null

Extra model settings inferred from namespace

sandbox.failover_models.parent.parent.deprecated_on

string | null

For models whose deprecation date is known (past or future), to show errors and deny service, or show warnings

sandbox.failover_models.parent.parent.owner

string | null

The owner of the model. Sometimes, for a provider/model combination, many instances exist, trained on different data

sandbox.failover_models.parent.parent.parent

object | null

sandbox.failover_models.parent.parent.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.failover_models.parent.parent.prompt_id

string | null

The ID of prompt, used for this model

sandbox.failover_models.parent.parent.provider

string | null

The provider for the model.

sandbox.failover_models.parent.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.failover_models.parent.prompt_id

string | null

The ID of prompt, used for this model

sandbox.failover_models.parent.provider

string | null

The provider for the model.

sandbox.failover_models.parent_id

string | null

The ID of parent, in case it's not a base model

sandbox.failover_models.prompt_id

string | null

The ID of prompt, used for this model

sandbox.failover_models.provider

string | null

The provider for the model.

sandbox.current_user_permissions

string[] | null

sandbox_status

enum<string> | null

required

Available options:

outdated

Getting Started

Spaces Guide

Developer Guide

API REFERENCE

COMMUNITY

PULZE ACADEMY

Update Space And Models

Authorizations

Body

Response