API

An API compatible with OpenAI and Anthropic

Chat Completions, Responses, Images and Messages on one key. This page has addresses, models, endpoints and errors; setup for Cursor, Claude Code, Codex and other IDEs is under Integrations.

Create a key

In your profile, API tab. The key starts with sk-slafy- and is shown once, so save it.

Choose a model

Model IDs are listed below and on the pricing page. For example, claude-sonnet-5-5.

Set the address and key

For the OpenAI format the address is https://slafy.co/v1, for Anthropic https://slafy.co. Pass the key as Bearer or x-api-key.

Base URL: https://slafy.co/v1
For the Anthropic SDK: https://slafy.co, the SDK adds /v1 itself
Key: Authorization: Bearer sk-slafy-… or x-api-key: sk-slafy-… header
Model list: GET /v1/models, prices on the Pricing page
Reasoning effort: reasoning_effort (reasoning.effort in Responses): low, medium, high, xhigh, max or ultra, the last one being the maximum for the selected model. If a model doesn't support a value, Slafy substitutes the closest one, and for models without effort settings it removes the parameter

OpenAI SDK

Python and Node.js, any model via Chat Completions

Python

from openai import OpenAI

client = OpenAI(base_url="https://slafy.co/v1", api_key="sk-slafy-your_key")
r = client.chat.completions.create(
    model="claude-sonnet-5-5",
    messages=[{"role": "user", "content": "Hello"}],
)
print(r.choices[0].message.content)

Node.js

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://slafy.co/v1", apiKey: process.env.SLAFY_API_KEY });
const r = await client.chat.completions.create({
  model: "gpt-6-luna",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(r.choices[0].message.content);

Anthropic SDK

Python and Node.js, Claude models via Messages

Python

import anthropic

client = anthropic.Anthropic(base_url="https://slafy.co", api_key="sk-slafy-your_key")
m = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}],
)
print(m.content[0].text)

Node.js

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({ baseURL: "https://slafy.co", apiKey: process.env.SLAFY_API_KEY });
const m = await client.messages.create({
  model: "claude-sonnet-5-5",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello" }],
});
console.log(m.content[0].text);

curl

Check your key and model from the terminal

Chat Completions

curl https://slafy.co/v1/chat/completions \
  -H "Authorization: Bearer sk-slafy-your_key" \
  -H "Content-Type: application/json" \
  -d '{"model": "claude-sonnet-5-5", "messages": [{"role": "user", "content": "Hello"}]}'

Messages, streaming

curl https://slafy.co/v1/messages \
  -H "x-api-key: sk-slafy-your_key" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{"model": "claude-sonnet-5-5", "max_tokens": 512, "stream": true,
       "messages": [{"role": "user", "content": "Tell me a joke"}]}'

Image generation

The model draws inside its answer or through the Images API

One image costs $0.199, which is 500,000 balance tokens. Errors and rejected requests are free.

Ask the model to draw

Add the {"type": "image_generation"} tool to a Responses request. The model decides when to draw and returns the image in an image_generation_call item (base64 PNG in the result field). This is how Codex and agents built on the OpenAI SDK draw.

curl https://slafy.co/v1/responses \
  -H "Authorization: Bearer sk-slafy-your_key" \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-6-astra", "input": "Draw an app icon: a blue whale, flat style",
       "tools": [{"type": "image_generation", "quality": "low"}]}'

GPT models can draw: gpt-6.1-sol, gpt-6.1-terra, gpt-6-astra, gpt-6-luna. The text tokens of the answer are added to the image price.

Images API

from openai import OpenAI
import base64

client = OpenAI(base_url="https://slafy.co/v1", api_key="sk-slafy-your_key")
r = client.images.generate(model="gpt-image-2.5-sunburst", prompt="A blue whale, flat style", size="1024x1024")
open("whale.png", "wb").write(base64.b64decode(r.data[0].b64_json))

The names gpt-image-2, gpt-image-1, dall-e-3 work too: Slafy sends them to gpt-image-2.5-sunburst. So ready-made scripts and skills written for OpenAI work without changes.

Models

IDs for the model field and the API formats each model works with. Prices are on the pricing page.

Anthropic

Claude Opus 5.5claude-opus-5-5chat/completions · responses · messages
Claude Sonnet 5.5claude-sonnet-5-5chat/completions · responses · messages
Claude Haiku 5.5claude-haiku-5-5chat/completions · responses · messages
Claude Fable 5.1claude-fable-5-1chat/completions · responses · messages

OpenAI

GPT-6.1 Solgpt-6.1-solchat/completions · responses · messages
GPT-6.1 Terragpt-6.1-terrachat/completions · responses
GPT-6 Astragpt-6-astrachat/completions · responses · messages
GPT-6 Lunagpt-6-lunachat/completions · responses · messages
GPT Image 2.5 Sunburstgpt-image-2.5-sunburstimages

Google

Gemini 4 Argongemini-4-argonchat/completions
Gemini 3.8 Flashgemini-3.8-flashchat/completions

xAI

Grok 4.7grok-4.7chat/completions · responses · messages

DeepSeek

DeepSeek V4 Prodeepseek-v4-prochat/completions
DeepSeek V4.1 Flash 1017deepseek-v4.1-flash-1017chat/completions

Moonshot AI

Kimi K3.1 Prokimi-k3.1-prochat/completions

Zhipu AI

GLM 5.3glm-5.3chat/completions · responses · messages
GLM 5.3 Flashglm-5.3-flashchat/completions · responses · messages

Endpoints

Method and path	Format
`POST /v1/chat/completions`	OpenAI Chat Completions, streaming supported
`POST /v1/responses`	OpenAI Responses, streaming and the `image_generation` tool
`POST /v1/messages`	Anthropic Messages, streaming supported
`POST /v1/messages/count_tokens`	Anthropic, token counting, free
`POST /v1/images/generations`	OpenAI Images, response in `b64_json`
`POST /v1/images/edits`	OpenAI Images, multipart/form-data
`GET /v1/models`	List of available models

Which model works with which endpoint is shown in pricing.

Errors

HTTP	Code	What to do
401	`invalid_api_key`	Key not found or disabled. Copy the full key from your profile.
402	`insufficient_balance`	You're out of tokens, top up your balance.
402	`key_limit_reached`	The key has hit its spending limit, raise the limit in the API tab.
404	`model_not_found`	No such model. Check the name with `GET /v1/models`.
429	`rate_limit_exceeded`	Too many requests. Try again in a few seconds.
5xx	`upstream_error`	Model-side failure, the request isn't charged. Try again.

Every response has a x-request-id header that you can use to find the request in your history.