Documentation

The API accepts the OpenAI Chat Completions and Anthropic Messages formats, so existing clients work with a different base URL and an API key.

Quickstart

A streamed request to deepseek-v4.1-flash-1017.

curl https://sableapi.com/v1/chat/completions \  -H "Authorization: Bearer $SABLE_API_KEY" \  -H "Content-Type: application/json" \  -d '{    "model": "deepseek-v4.1-flash-1017",    "stream": true,    "messages": [      {"role": "user", "content": "List the regressions in this diff."}    ]  }'

from openai import OpenAI client = OpenAI(base_url="https://sableapi.com/v1", api_key=SABLE_API_KEY) stream = client.chat.completions.create(    model="deepseek-v4.1-flash-1017",    stream=True,    messages=[        {"role": "user", "content": "List the regressions in this diff."}    ],)for chunk in stream:    print(chunk.choices[0].delta.content or "", end="")

curl https://sableapi.com/v1/messages \  -H "Authorization: Bearer $SABLE_API_KEY" \  -H "Content-Type: application/json" \  -d '{    "model": "deepseek-v4.1-flash-1017",    "max_tokens": 1024,    "stream": true,    "messages": [      {"role": "user", "content": "Summarise the open questions in this thread."}    ]  }'

Base URL and authentication

Base URL: https://sableapi.com/v1
Header: Authorization: Bearer <key>
Keys: Created and revoked in your account.

Endpoints

Method and path	Description
`POST /v1/chat/completions`	OpenAI Chat Completions format. Streams with `"stream": true`.
`POST /v1/messages`	Anthropic Messages format. Streams with `"stream": true`.
`GET /v1/models`	Models available to the key.
`GET /v1/requests/{id}`	The record of one request.

Parameters

Sampling parameters are passed to the model as sent: temperature, top_p, top_k, min_p, frequency_penalty, presence_penalty, seed, stop and logit_bias.

Errors

Errors use the error format of the endpoint that was called, with a code field. A key may run any number of requests at once.

Status	Code	Meaning
`400`	`invalid_request`	The request body is not valid.
`400`	`context_length_exceeded`	The input is longer than 512,000 tokens.
`401`	`invalid_api_key`	The API key is missing or not valid.
`402`	`insufficient_credits`	The balance is too low for the request.
`404`	`model_not_found`	The model is not available.
`409`	`idempotency_conflict`	The idempotency key was already used for a different request.
`503`	`capacity_unavailable`	All capacity is in use. Retry after the `Retry-After` header; reuse the idempotency key.
`503`	`service_unavailable`	The service is temporarily unavailable. Retry after the `Retry-After` header.