API Reference

Genfity AI Gateway provides a single endpoint compatible with standard chat completions (OpenAI) and Messages API (Anthropic) formats to access all AI models. Just change the base URL and use your Genfity API key.

BASE URL (OpenAI)	https://ai.genfity.com/v1
BASE URL (Anthropic)	https://ai.genfity.com

Quick start

Quick startcurl

curl https://ai.genfity.com/v1/models \
  -H "Authorization: Bearer genfity_xxxxxxxxxxxx"

Authentication

All requests require an API key sent via the Authorization header. Get your API key at Dashboard → API Keys.

Header	Type	Description
Authorization	string required	API Key in Bearer <api_key> format. Get it from the dashboard.

Examplebash

curl https://ai.genfity.com/v1/chat/completions \
  -H "Authorization: Bearer genfity_xxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{"model":"genfity/claude-opus-4.7","messages":[{"role":"user","content":"Hello"}]}'

Endpoints

Send requests to any AI model through the following endpoints.

POST/v1/chat/completionsOpenAI-compatible chat completion

POST/v1/messagesAnthropic-compatible messages

POST/v1/messages/count_tokensToken count estimation

POST/v1/embeddingsVector embeddings

GET/v1/modelsList available models

POST

Create Chat Completion (OpenAI)

/v1/chat/completions

Send an OpenAI-compatible chat completion request to any AI model in the Genfity catalog.

Headers

Authorization	string required	API Key in Bearer <api_key> format.
Content-Type	string required	Must be application/json.

Request Body

model	string required	Model ID from catalog. See /v1/models.
messages	array required	Array of message objects with role and content.
stream	boolean	Enable SSE streaming. Default false.
max_tokens	integer	Maximum output tokens.
temperature	number	Sampling temperature (0-2).
top_p	number	Nucleus sampling (0-1).
tools	array	OpenAI-format tool calling definitions.
response_format	object	Output format, e.g. { type: json_object }.

Requestcurl

curl -X POST https://ai.genfity.com/v1/chat/completions \
  -H "Authorization: Bearer genfity_xxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "genfity/claude-opus-4.7",
    "messages": [
      { "role": "user", "content": "Apa ibukota Indonesia?" }
    ],
    "temperature": 0.7,
    "max_tokens": 256,
    "stream": false
  }'

Responsejson

{
  "id": "chatcmpl_01Hxx...",
  "object": "chat.completion",
  "created": 1778803200,
  "model": "genfity/claude-opus-4.7",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Ibukota Indonesia adalah Jakarta."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 14,
    "completion_tokens": 8,
    "total_tokens": 22
  }
}

POST

Anthropic Messages

/v1/messages

Send an Anthropic-compatible request to any AI model. Suitable for Claude SDKs and tools requiring Messages API.

Headers

Authorization	string required	API Key in Bearer <api_key> format.
Content-Type	string required	Must be application/json.

Request Body

model	string required	Model ID from catalog.
max_tokens	integer required	Maximum output tokens.
messages	array required	Array of messages with role and content.
system	string	System prompt (optional).
stream	boolean	Enable SSE streaming.
tools	array	Tool definitions with input_schema.
tool_choice	object	Control which tool is called.

Requestcurl

curl -X POST https://ai.genfity.com/v1/messages \
  -H "Authorization: Bearer genfity_xxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "genfity/claude-opus-4.7",
    "max_tokens": 1024,
    "messages": [
      { "role": "user", "content": "Hello Genfity" }
    ]
  }'

Responsejson

{
  "id": "msg_01Hxx...",
  "type": "message",
  "role": "assistant",
  "model": "genfity/claude-opus-4.7",
  "content": [
    { "type": "text", "text": "Hello! How can I help you today?" }
  ],
  "stop_reason": "end_turn",
  "usage": {
    "input_tokens": 25,
    "output_tokens": 12
  }
}

POST

Count Tokens

/v1/messages/count_tokens

Estimate the number of tokens in a prompt before sending a generation request.

Headers

Authorization

string required

API Key in Bearer <api_key> format.

Request Body

model	string required	Model ID from catalog.
messages	array required	Messages to count.

Requestcurl

curl -X POST https://ai.genfity.com/v1/messages/count_tokens \
  -H "Authorization: Bearer genfity_xxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "genfity/claude-opus-4.7",
    "messages": [{"role":"user","content":"Count this text"}]
  }'

Responsejson

{
  "input_tokens": 12
}

POST

Embeddings

/v1/embeddings

Generate vector embeddings from input text. Only available for embedding models in the catalog.

Headers

Authorization

string required

API Key in Bearer <api_key> format.

Request Body

model	string required	Embedding model ID from catalog.
input	string \| array required	Text or array of texts to embed.

Requestcurl

curl -X POST https://ai.genfity.com/v1/embeddings \
  -H "Authorization: Bearer genfity_xxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "genfity/claude-opus-4.7",
    "input": "Genfity AI Gateway makes AI integration simple."
  }'

Responsejson

{
  "object": "list",
  "data": [
    {
      "object": "embedding",
      "index": 0,
      "embedding": [0.0023, -0.0091, 0.0042, ...]
    }
  ],
  "model": "genfity/claude-opus-4.7",
  "usage": { "prompt_tokens": 8, "total_tokens": 8 }
}

POST

Image Generation

/v1/images/generations

Generate images from text prompts using image generation models like GPT Image, FLUX, Stable Diffusion, and more.

Headers

Authorization	string required	API Key in Bearer <api_key> format.
Content-Type	string required	Must be application/json.

Request Body

model	string required	Image model ID, e.g. openai/gpt-image-2, together/black-forest-labs/FLUX.2-pro, flux-kontext.
prompt	string required	Description of the image to generate.
n	integer	Number of images (default 1).
size	string	Image size, e.g. 1024x1024, 1024x1792.
quality	string	Quality: standard or hd.
response_format	string	Output format: url or b64_json.

Requestcurl

curl -X POST https://ai.genfity.com/v1/images/generations \
  -H "Authorization: Bearer genfity_xxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-image-2",
    "prompt": "A beautiful sunset over Indonesian rice terraces",
    "n": 1,
    "size": "1024x1024"
  }'

Responsejson

{
  "created": 1778803200,
  "data": [
    {
      "url": "https://...",
      "revised_prompt": "A beautiful sunset over Indonesian rice terraces"
    }
  ]
}

GET

List Models

/v1/models

Get a list of all models available to your account.

Headers

Authorization

string required

API Key in Bearer <api_key> format.

Requestcurl

curl https://ai.genfity.com/v1/models \
  -H "Authorization: Bearer genfity_xxxxxxxxxxxx"

Responsejson

{
  "object": "list",
  "data": [
    { "id": "genfity/claude-opus-4.7", "object": "model", "owned_by": "genfity" }
  ]
}

Model & Pricing

Supported models and pricing per 1M tokens. Data is fetched directly from the active catalog database.

Loading models from database...

Error Codes

Genfity AI Gateway uses standard HTTP status codes.

Code	Message	Meaning
400	`invalid_json`	Body is not valid JSON.
400	`missing_model`	Model field is empty.
400	`model_not_allowed`	Model not found or inactive in catalog.
400	`max_tokens_required`	PAYG/credit requires max_tokens limit.
401	`missing_api_key`	Authorization header is empty.
401	`invalid_api_key_format`	Header is not a Bearer token.
401	`invalid_api_key`	Key is invalid or revoked.
402	`no_active_subscription`	Key has no active subscription.
402	`subscription_not_covering_model`	Plan does not cover this model.
402	`credit_cost_not_configured`	Model has no credit cost configured.
402	`payg_price_not_configured`	Model has no PAYG price configured.
402	`insufficient_credit_balance`	Insufficient credit balance.
402	`insufficient_balance`	Insufficient balance.
403	`forbidden`	No access to this model.
429	`rate_limit_exceeded`	RPM/TPM/concurrency limit exceeded.
429	`quota_exceeded`	Period token quota exhausted.
429	`global_rate_limit_exceeded`	IP burst guard exceeded.
500	`internal_error`	Internal gateway error.
502	`router_unavailable`	The model/provider is under high traffic. Try again later.
502	`upstream_error`	The model/provider is under high traffic. Try again later.
502	`all_candidates_failed`	The model/provider is under high traffic. Try again later.
503	`service_unavailable`	Service is under maintenance.

Rate Limits

Request limits are set per API key from Dashboard → API Keys. When limits are exceeded, the API returns 429 Too Many Requests.

Per-key limits

• rate_limit_per_min — Requests per minute per key
• quota_tokens_monthly — Monthly token quota
• concurrent_limit — Concurrent requests

Global IP guard

• Window: 1 minute
• Scope: IP + endpoint path
• Error: global_rate_limit_exceeded

IDE & Editor

Genfity AI Gateway uses OpenAI-compatible and Anthropic-compatible formats, so almost any IDE, CLI, or SDK that supports custom base URLs works out of the box. Just change the base URL and use your Dashboard API key.

Cline

Autonomous coding agent for VS Code

Cline supports OpenAI Compatible providers. Fill in base URL, API key, and model ID in the Settings panel.

Setup

1Install the Cline extension from VS Code Marketplace.
2Open the Cline panel (left sidebar), click the Settings icon.
3API Provider → choose OpenAI Compatible.
4Fill the fields according to the snippet, click Done.

https://cline.bot

Clineini

API Provider   = OpenAI Compatible
Base URL       = https://ai.genfity.com/v1
API Key        = genfity_xxxxxxxxxxxx
Model ID       = genfity/claude-opus-4.7

Roo Code

Cline fork with multi-mode agents

Roo Code (Cline fork) uses the same configuration — choose OpenAI Compatible then fill the Genfity base URL and key.

Setup

1Install the Roo Code extension from VS Code Marketplace.
2Open the Roo Code panel, click Settings or profile name.
3API Provider: OpenAI Compatible (not OpenAI).
4Save the profile, pick a mode (Code/Architect/Ask) and start prompting.

https://roocode.com

Roo Codeini

API Provider   = OpenAI Compatible
Base URL       = https://ai.genfity.com/v1
API Key        = genfity_xxxxxxxxxxxx
Model ID       = genfity/claude-opus-4.7

Continue.dev

Autopilot for VS Code & JetBrains

Continue uses ~/.continue/config.yaml to declare providers. Add an openai-compatible entry with the Genfity apiBase.

Setup

1Install the Continue extension in VS Code / JetBrains.
2Open ~/.continue/config.yaml (Continue: Open Config).
3Add a Genfity model under the models key.
4Reload — the model will appear in Continue's chat dropdown.

https://continue.dev

Continue.devyaml

models:
  - name: Genfity Claude Opus
    provider: openai
    model: genfity/claude-opus-4.7
    apiBase: https://ai.genfity.com/v1
    apiKey: genfity_xxxxxxxxxxxx
    roles:
      - chat
      - edit
      - apply

Cursor

AI-native code editor

Cursor supports Override OpenAI Base URL at Settings → Models. Enable OpenAI API Key, then point base URL to Genfity.

Setup

1Cursor Settings → Models → scroll to OpenAI API Key.
2Enter API key: genfity_xxxxxxxxxxxx.
3Expand section → enable Override OpenAI Base URL.
4Set base URL: https://ai.genfity.com/v1. Click Verify.
5Enable only models available in the Genfity catalog.

https://cursor.com

Cursorini

OpenAI API Key           = genfity_xxxxxxxxxxxx
Override Base URL        = https://ai.genfity.com/v1
Enabled Models           = genfity/claude-opus-4.7, ...

Windsurf (Codeium)

Agentic IDE by Codeium

Windsurf Cascade supports custom model endpoints. Add an OpenAI-compatible provider via Settings → Cascade → Add Custom Model.

Setup

1Settings (Cmd/Ctrl + ,) → Cascade → Models.
2Add Custom Model → choose OpenAI protocol.
3Fill Base URL + Genfity API Key + Model ID.
4Enable the new model and start a Cascade session.

https://windsurf.com

Windsurf (Codeium)ini

Protocol     = OpenAI
Endpoint     = https://ai.genfity.com/v1
API Key      = genfity_xxxxxxxxxxxx
Model ID     = genfity/claude-opus-4.7
Display Name = Genfity Claude Opus

Zed

High-performance collaborative editor

Zed supports OpenAI API Compatible providers via settings.json. Add Genfity as a custom provider, then store the API key through Agent Panel settings.

Setup

1Zed → Cmd/Ctrl + , → open settings.json.
2Add the language_models.openai_compatible key as in the snippet.
3Open Agent Panel settings (Cmd/Ctrl + Shift + P → agent: open settings).
4Find the "Genfity AI Gateway" provider, paste your API key, press Enter.
5Select a Genfity model from the Agent Panel model dropdown.

https://zed.dev

Zedjson

{
  "language_models": {
    "openai_compatible": {
      "Genfity AI Gateway": {
        "api_url": "https://ai.genfity.com/v1",
        "available_models": [
          {
            "name": "genfity/auto",
            "display_name": "Genfity Auto (Combo)",
            "max_tokens": 200000,
            "capabilities": {
              "tools": true,
              "images": true
            }
          },
          {
            "name": "genfity/claude-opus-4.7",
            "display_name": "Genfity Claude Opus 4.7",
            "max_tokens": 200000,
            "capabilities": {
              "tools": true,
              "images": true
            }
          },
          {
            "name": "genfity/claude-haiku-4.5",
            "display_name": "Genfity Claude Haiku 4.5",
            "max_tokens": 200000,
            "capabilities": {
              "tools": true,
              "images": true
            }
          }
        ]
      }
    }
  },
  "agent": {
    "default_model": {
      "provider": "openai_compatible",
      "model": "genfity/auto"
    }
  }
}

Kilo Code

Open-source AI agent for VS Code

Kilo Code supports OpenAI Compatible providers. Fill in the Genfity base URL, API key, and model ID in the Settings panel.

Setup

1Install the Kilo Code extension from VS Code Marketplace.
2Open the Kilo Code panel → Settings.
3API Provider → choose OpenAI Compatible.
4Fill Base URL + API Key + Model, click Save then Done.

https://kilocode.ai

Kilo Codeini

API Provider   = OpenAI Compatible
Base URL       = https://ai.genfity.com/v1
API Key        = genfity_xxxxxxxxxxxx
Model          = genfity/claude-opus-4.7

TRAE

ByteDance AI-native IDE

TRAE can add custom OpenAI-compatible models. Use the OpenAI protocol with the Genfity base URL and key.

Setup

1Open Settings (icon at top-right of the chat) → Models.
2Click + Add Model → choose the OpenAI Compatible protocol/provider.
3Fill Base URL = https://ai.genfity.com/v1, API Key, and Model = genfity/claude-opus-4.7.
4Click Add Model then select it in the chat.

https://trae.ai

TRAEini

Provider   = OpenAI Compatible
Base URL   = https://ai.genfity.com/v1
API Key    = genfity_xxxxxxxxxxxx
Model      = genfity/claude-opus-4.7

Xcode

Apple's IDE with Coding Intelligence

Xcode 26+ supports Internet Hosted model providers. Use the Genfity host with an Authorization: Bearer header.

Setup

1Xcode → Settings → Intelligence → Add a Model Provider.
2Choose the Internet Hosted tab.
3URL = https://ai.genfity.com (bare host, no /v1). API Key Header = Authorization.
4API Key = Bearer genfity_xxxxxxxxxxxx (the word Bearer, a space, then the key).
5Click Add, enable genfity/claude-opus-4.7, press Cmd+0 to start.

https://developer.apple.com/xcode/

Xcodeini

URL             = https://ai.genfity.com
API Key Header  = Authorization
API Key         = Bearer genfity_xxxxxxxxxxxx
Description      = Genfity AI Gateway
Model            = genfity/claude-opus-4.7

CLI & Terminal

Claude Code

Anthropic's official CLI

Claude Code uses the Anthropic-compatible endpoint. The most stable setup is an env block in ~/.claude/settings.json (not one-off exports). Clear any old ANTHROPIC_AUTH_TOKEN & ANTHROPIC_BASE_URL first — env vars override settings.json.

Prerequisites: Node.js 18+. On Windows prefer WSL or Git for Windows; on macOS use nvm (avoid a direct package install to dodge permission issues).

Setup

1IMPORTANT — clear old env: environment variables override settings.json. macOS/Linux: unset ANTHROPIC_AUTH_TOKEN ANTHROPIC_BASE_URL then delete the export lines in ~/.bashrc or ~/.zshrc. Windows PowerShell: [Environment]::SetEnvironmentVariable('ANTHROPIC_BASE_URL',$null,'User').
2Edit or create ~/.claude/settings.json (Windows: C:\Users\<user>\.claude\settings.json) with the env block in the Configuration panel.
3(Optional) Skip Anthropic login: set "hasCompletedOnboarding": true in ~/.claude.json so it won't try to reach api.anthropic.com.
4Open a NEW terminal (so config is read), cd into the project, run claude, choose Trust This Folder. Confirm with /status that ANTHROPIC_BASE_URL points to Genfity.

https://claude.com/claude-code

Install

bash

npm install -g @anthropic-ai/claude-code
claude --version

Configuration

Claude Codejson

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://ai.genfity.com",
    "ANTHROPIC_AUTH_TOKEN": "genfity_xxxxxxxxxxxx",
    "ANTHROPIC_MODEL": "genfity/claude-opus-4.7",
    "ANTHROPIC_DEFAULT_OPUS_MODEL": "genfity/claude-opus-4.7",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "genfity/claude-opus-4.7",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "genfity/claude-haiku-4.5",
    "ANTHROPIC_SMALL_FAST_MODEL": "genfity/claude-haiku-4.5",
    "API_TIMEOUT_MS": "3000000"
  }
}

opencode

AI coding agent in your terminal (SST)

opencode is an open-source TUI-based coding agent. Add Genfity as an OpenAI-compatible provider.

Setup

1Create ~/.config/opencode/opencode.json (global) or opencode.json at project root (Windows: %USERPROFILE%\.config\opencode\opencode.json).
2Set the API key as env GENFITY_KEY. macOS/Linux: export GENFITY_KEY="KEY" in ~/.zshrc/~/.bashrc. Windows PowerShell: [Environment]::SetEnvironmentVariable('GENFITY_KEY','KEY','User').
3Run opencode in your terminal, pick the Genfity provider from the model list.

https://opencode.ai

Install

bash

curl -fsSL https://opencode.ai/install | bash

Configuration

opencodejson

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "genfity": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Genfity",
      "options": {
        "baseURL": "https://ai.genfity.com/v1",
        "apiKey": "{env:GENFITY_KEY}"
      },
      "models": {
        "genfity/claude-opus-4.7": { "name": "Claude Opus 4.7" }
      }
    }
  }
}

Codex CLI

OpenAI Codex CLI agent

Codex CLI supports OpenAI-compatible providers via ~/.codex/config.toml. The Genfity gateway serves Chat/Completions, so wire_api = "chat" is required — the latest Codex defaults to the Responses API, which the gateway does not serve.

Prerequisites: Node.js 18+. The Genfity gateway only serves Chat/Completions, so wire_api = "chat" is REQUIRED — the latest Codex defaults to the unsupported Responses API.

Setup

1Set the API key as the GENFITY_KEY environment variable (see the Set API key panel).
2Edit ~/.codex/config.toml (Windows: C:\Users\<user>\.codex\config.toml) — add the block in the Configuration panel. wire_api = "chat" is REQUIRED.
3If the latest Codex throws "wire_api = chat is no longer supported", downgrade: npm install -g @openai/[email protected].
4Restart Codex, then run codex at the project root (model & provider are read from config automatically).

https://github.com/openai/codex

Install

bash

npm i -g @openai/codex
codex --version

Configuration

Codex CLItoml

model = "genfity/claude-opus-4.7"
model_provider = "genfity"
model_context_window = 200000

[model_providers.genfity]
name = "Genfity"
base_url = "https://ai.genfity.com/v1"
env_key = "GENFITY_KEY"
wire_api = "chat"

Set API key (GENFITY_KEY)

Set API key (GENFITY_KEY)bash

# macOS / Linux (zsh)
echo 'export GENFITY_KEY="genfity_xxxxxxxxxxxx"' >> ~/.zshrc
source ~/.zshrc

# Windows PowerShell
[Environment]::SetEnvironmentVariable('GENFITY_KEY','genfity_xxxxxxxxxxxx','User')

Windows: open a new PowerShell after setting the env var so it's picked up.

Aider

AI pair programming in your terminal

Aider reads OpenAI-style environment variables. Override OPENAI_API_BASE and OPENAI_API_KEY for Genfity.

Prerequisites: Python 3.8+. The openai/ model prefix is required so Aider routes to the OpenAI-compatible endpoint.

Setup

1Set OpenAI-style env (see the Configuration panel). Or put them in .aider.conf.yml / .env. Windows PowerShell: [Environment]::SetEnvironmentVariable('OPENAI_API_BASE','...','User') and the same for OPENAI_API_KEY.
2Run from project root: aider --model openai/genfity/claude-opus-4.7.

https://aider.chat

Install

bash

pipx install aider-chat

Configuration

Aiderbash

export OPENAI_API_BASE="https://ai.genfity.com/v1"
export OPENAI_API_KEY="genfity_xxxxxxxxxxxx"

aider --model openai/genfity/claude-opus-4.7

LLM (Simon Willison)

Versatile CLI + Python library for any LLM

LLM CLI uses the llm-openai-plugin for OpenAI-compatible providers. Register Genfity as a new model.

Prerequisites: Python 3.8+.

Setup

1Save API key: llm keys set genfity then paste genfity_xxxxxxxxxxxx.
2Find the config dir: dirname "$(llm logs path)". Edit extra-openai-models.yaml there (create it if missing) as in the Configuration panel.
3Test: llm -m genfity/claude-opus 'hello'.

https://llm.datasette.io

Install

bash

pipx install llm
llm install llm-openai-plugin

Configuration

LLM (Simon Willison)yaml

- model_id: genfity/claude-opus
  model_name: genfity/claude-opus-4.7
  api_base: "https://ai.genfity.com/v1"
  api_key_name: genfity
  can_stream: true

Goose

Block's agentic CLI + desktop

Goose supports Anthropic Compatible providers. Point the Base URL to the Genfity Anthropic endpoint (without /v1).

Setup

1Open Goose → Create New Provider → choose Anthropic Compatible.
2Fill Base URL = https://ai.genfity.com (no /v1), API Key = genfity_xxxxxxxxxxxx, Model = genfity/claude-opus-4.7.
3Save, then Switch Models on the main screen and pick the Genfity provider.

https://block.github.io/goose/

Install

bash

curl -fsSL https://github.com/block/goose/releases/download/stable/download_cli.sh | bash

Configuration

Gooseini

Provider  = Anthropic Compatible
Base URL  = https://ai.genfity.com
API Key   = genfity_xxxxxxxxxxxx
Model     = genfity/claude-opus-4.7

Crush

Charm's terminal coding agent (CLI + TUI)

Crush uses ~/.config/crush/crush.json to declare OpenAI-compatible providers. Add Genfity under the providers key.

Setup

1Edit ~/.config/crush/crush.json (Windows: %USERPROFILE%\.config\crush\crush.json). Create the folder if missing.
2Add the providers.genfity block as in the Configuration panel.
3Restart crush, press Ctrl+P → Switch Model → pick a Genfity model.

https://github.com/charmbracelet/crush

Install

bash

brew install charmbracelet/tap/crush

Configuration

Crushjson

{
  "providers": {
    "genfity": {
      "id": "genfity",
      "name": "Genfity AI Gateway",
      "base_url": "https://ai.genfity.com/v1",
      "type": "openai",
      "api_key": "genfity_xxxxxxxxxxxx",
      "models": [
        { "id": "genfity/claude-opus-4.7", "name": "Genfity Claude Opus 4.7" }
      ]
    }
  }
}

Droid (Factory)

Factory's terminal coding agent

Droid uses ~/.factory/config.json for custom models. Use the anthropic provider with the Genfity base URL.

Prerequisites: Clear ANTHROPIC_AUTH_TOKEN — it overrides api_key in config.json. macOS/Linux: unset ANTHROPIC_AUTH_TOKEN. Windows PowerShell: [Environment]::SetEnvironmentVariable('ANTHROPIC_AUTH_TOKEN',$null,'User').

Setup

1Edit ~/.factory/config.json (not settings.json; Windows: C:\Users\<user>\.factory\config.json) — add the custom_models entry from the Configuration panel.
2Run droid, type /model then pick Genfity.

https://factory.ai

Install

bash

curl -fsSL https://app.factory.ai/cli | sh

Configuration

Droid (Factory)json

{
  "custom_models": [
    {
      "model_display_name": "Genfity Claude Opus 4.7",
      "model": "genfity/claude-opus-4.7",
      "base_url": "https://ai.genfity.com",
      "api_key": "genfity_xxxxxxxxxxxx",
      "provider": "anthropic",
      "max_tokens": 64000
    }
  ]
}

Qwen Code

Alibaba's terminal coding agent

Qwen Code reads OpenAI-style environment variables. Override OPENAI_BASE_URL, OPENAI_API_KEY, and OPENAI_MODEL for Genfity.

Prerequisites: Node.js 20+.

Setup

1Set OpenAI-style env (see the Configuration panel). Windows PowerShell: [Environment]::SetEnvironmentVariable('OPENAI_BASE_URL','...','User') and the same for OPENAI_API_KEY / OPENAI_MODEL.
2Open a new terminal, run qwen at the project root. If prompted for a provider, pick OpenAI-compatible and confirm the model ID.

https://github.com/QwenLM/qwen-code

Install

bash

npm i -g @qwen-code/qwen-code

Configuration

Qwen Codebash

export OPENAI_BASE_URL="https://ai.genfity.com/v1"
export OPENAI_API_KEY="genfity_xxxxxxxxxxxx"
export OPENAI_MODEL="genfity/claude-opus-4.7"

qwen

Grok CLI

Terminal agent (OpenAI-compatible)

Grok CLI uses GROK_BASE_URL and GROK_API_KEY. Point them at the Genfity OpenAI-compatible endpoint.

Prerequisites: Node.js 18+. Clear OPENAI_API_KEY & OPENAI_BASE_URL to avoid clashes. macOS/Linux: unset OPENAI_API_KEY OPENAI_BASE_URL. Windows PowerShell: [Environment]::SetEnvironmentVariable('OPENAI_API_KEY',$null,'User').

Setup

1Set Grok env (see the Configuration panel). Windows PowerShell: [Environment]::SetEnvironmentVariable('GROK_BASE_URL','...','User') and the same for GROK_API_KEY.
2Open a new terminal, run grok --model genfity/claude-opus-4.7.

https://github.com/superagent-ai/grok-cli

Install

bash

npm i -g @vibe-kit/grok-cli

Configuration

Grok CLIbash

export GROK_BASE_URL="https://ai.genfity.com/v1"
export GROK_API_KEY="genfity_xxxxxxxxxxxx"

grok --model genfity/claude-opus-4.7

OpenHands

Open-source AI software agent

OpenHands uses the openai/ prefix for OpenAI-compatible endpoints. Set llm_base_url to Genfity.

Prerequisites: Python 3.12+. The model MUST carry the openai/ prefix. On Windows use WSL.

Setup

1Configure the LLM via env (LLM_MODEL/LLM_BASE_URL/LLM_API_KEY) or config.toml — see the Configuration panel.
2Run openhands then open the UI at http://localhost:3000.

https://github.com/All-Hands-AI/OpenHands

Install

bash

pipx install openhands

Configuration

OpenHandsini

llm_model="openai/genfity/claude-opus-4.7"
llm_base_url="https://ai.genfity.com/v1"
llm_api_key="genfity_xxxxxxxxxxxx"

openhands

OpenClaw

Open-source personal AI assistant (WA/Telegram/Discord)

OpenClaw uses an anthropic-messages provider in ~/.openclaw/openclaw.json. Register Genfity as a provider then restart the gateway.

Prerequisites: Node.js 22.19+. auth.mode none is single-machine local only; for remote access run openclaw doctor --fix.

Setup

1Run the wizard: openclaw onboard. For the provider choose Skip for now (registered manually next).
2Edit ~/.openclaw/openclaw.json (Windows: %USERPROFILE%\.openclaw\openclaw.json) — add the genfity provider from the Configuration panel. Use mode: merge to keep other config safe; don't replace the whole file.
3Set agents.defaults.model.primary to genfity/claude-opus-4.7.
4Apply: openclaw gateway restart. WhatsApp/Telegram/Discord channels are configured separately after the gateway is up.

https://openclaw.ai

Install

bash

curl -fsSL https://openclaw.ai/install.sh | bash

Configuration

OpenClawjson

{
  "models": {
    "mode": "merge",
    "providers": {
      "genfity": {
        "baseUrl": "https://ai.genfity.com",
        "apiKey": "genfity_xxxxxxxxxxxx",
        "api": "anthropic-messages",
        "models": [
          {
            "id": "genfity/claude-opus-4.7",
            "name": "Genfity Claude Opus 4.7",
            "input": ["text", "image"],
            "contextWindow": 200000,
            "maxTokens": 64000
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": { "primary": "genfity/claude-opus-4.7" }
    }
  }
}

Desktop & Web UI

LibreChat

Self-hosted multi-model chat UI

LibreChat uses librechat.yaml to register custom endpoints. Add Genfity as a custom endpoint.

Setup

1Copy librechat.example.yaml → librechat.yaml (deploy root).
2Add the endpoints.custom block as shown.
3Set GENFITY_KEY in .env (mounted to the container).
4Restart the container: docker compose up -d.

https://librechat.ai

LibreChatyaml

version: 1.0.9
endpoints:
  custom:
    - name: "Genfity"
      apiKey: "${GENFITY_KEY}"
      baseURL: "https://ai.genfity.com/v1"
      models:
        default: ["genfity/claude-opus-4.7"]
        fetch: true
      titleModel: "genfity/claude-haiku-4.5"
      summarize: false
      forcePrompt: false

Open WebUI

Extensible ChatGPT-style web UI

Open WebUI can add multiple OpenAI-compatible connections in Admin Settings.

Setup

1Sign in as admin → Admin Panel → Settings → Connections.
2OpenAI API → click + to add a new connection.
3Fill: API Base URL = https://ai.genfity.com/v1, API Key = genfity_xxxxxxxxxxxx.
4Save → Genfity models appear in the Workspace dropdown.

https://openwebui.com

Open WebUIini

API Base URL = https://ai.genfity.com/v1
API Key      = genfity_xxxxxxxxxxxx
Prefix ID    = genfity (optional)

Chatbox

Cross-platform desktop AI client

Chatbox supports a custom OpenAI host. Open Settings → AI Model → choose OpenAI API and fill the Genfity host + key.

Setup

1Download Chatbox (macOS/Windows/Linux) or use the web version.
2Settings → Model → AI Provider = OpenAI API.
3Fill API Key + API Host = https://ai.genfity.com/v1.
4Choose Model = genfity/claude-opus-4.7 → Save.

https://chatboxai.app

Chatboxini

AI Provider = OpenAI API
API Key     = genfity_xxxxxxxxxxxx
API Host    = https://ai.genfity.com/v1
Model       = genfity/claude-opus-4.7

Jan

Offline-first personal AI

Jan supports remote providers via Settings → Model Providers → OpenAI.

Setup

1Settings → Model Providers → OpenAI.
2Fill API Key = genfity_xxxxxxxxxxxx, Base URL = https://ai.genfity.com/v1.
3Models tab → import a model with ID genfity/claude-opus-4.7.
4Start a new thread and pick the Genfity model.

https://jan.ai

Janini

Provider  = OpenAI (compatible)
Base URL  = https://ai.genfity.com/v1
API Key   = genfity_xxxxxxxxxxxx
Model ID  = genfity/claude-opus-4.7

LM Studio

Desktop UI for local + remote LLMs

LM Studio supports remote OpenAI-compatible endpoints via the Connect to a server feature.

Setup

1LM Studio → Chat icon → click Select a model.
2Dropdown → Server settings → Add remote endpoint.
3Base URL = https://ai.genfity.com/v1, API Key = genfity_xxxxxxxxxxxx.
4Click Load → choose genfity/claude-opus-4.7.

https://lmstudio.ai

LM Studioini

Base URL = https://ai.genfity.com/v1
API Key  = genfity_xxxxxxxxxxxx
Model    = genfity/claude-opus-4.7

Cherry Studio

Multi-provider desktop AI client

Cherry Studio can add custom OpenAI-compatible providers. Fill the Genfity API host + key, then add the model.

Setup

1Settings → Model Providers → Add Provider → choose the OpenAI type.
2Fill API Host = https://ai.genfity.com/v1, API Key = genfity_xxxxxxxxxxxx.
3Click Check to verify, then Add Model = genfity/claude-opus-4.7.
4Pick the Genfity model in the chat dropdown.

https://cherry-ai.com

Cherry Studioini

Provider  = OpenAI (compatible)
API Host  = https://ai.genfity.com/v1
API Key   = genfity_xxxxxxxxxxxx
Model     = genfity/claude-opus-4.7

SDK & Framework

OpenAI Python SDK

Official Python client

The OpenAI Python SDK accepts a base_url argument. Just change the base URL and API key.

Setup

1Install: pip install openai.
2Instantiate the client with the Genfity base URL.
3Call chat.completions.create(...) as usual.

https://github.com/openai/openai-python

OpenAI Python SDKpython

from openai import OpenAI

client = OpenAI(
    api_key="genfity_xxxxxxxxxxxx",
    base_url="https://ai.genfity.com/v1",
)

res = client.chat.completions.create(
    model="genfity/claude-opus-4.7",
    messages=[{"role": "user", "content": "Hello Genfity"}],
)
print(res.choices[0].message.content)

OpenAI Node SDK

Official TypeScript / Node client

The OpenAI Node SDK accepts a baseURL on instantiation. Fully compatible with Genfity chat completions.

Setup

1Install: npm i openai.
2Set GENFITY_KEY env var, then use as in the snippet.

https://github.com/openai/openai-node

OpenAI Node SDKtypescript

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.GENFITY_KEY,
  baseURL: "https://ai.genfity.com/v1",
});

const res = await client.chat.completions.create({
  model: "genfity/claude-opus-4.7",
  messages: [{ role: "user", content: "Hello Genfity" }],
});
console.log(res.choices[0].message.content);

Anthropic Python SDK

Official Anthropic Python client

The Anthropic Python SDK accepts a base_url argument. Use the Genfity /v1/messages endpoint.

Setup

1Install: pip install anthropic.
2Instantiate the client with the Genfity base URL (without /v1).
3Call messages.create(...) as usual.

https://github.com/anthropics/anthropic-sdk-python

Anthropic Python SDKpython

import anthropic

client = anthropic.Anthropic(
    api_key="genfity_xxxxxxxxxxxx",
    base_url="https://ai.genfity.com",
)

message = client.messages.create(
    model="genfity/claude-opus-4.7",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello Genfity"}],
)
print(message.content[0].text)

Anthropic Node SDK

Official Anthropic TypeScript client

The Anthropic Node SDK accepts a baseURL on instantiation. Use the Genfity /v1/messages endpoint.

Setup

1Install: npm i @anthropic-ai/sdk.
2Instantiate the client with the Genfity base URL (without /v1).
3Call messages.create(...) as usual.

https://github.com/anthropics/anthropic-sdk-typescript

Anthropic Node SDKtypescript

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

const client = new Anthropic({
  apiKey: process.env.GENFITY_KEY,
  baseURL: "https://ai.genfity.com",
});

const message = await client.messages.create({
  model: "genfity/claude-opus-4.7",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello Genfity" }],
});
console.log(message.content[0].text);

LangChain

LLM orchestration framework

Use ChatOpenAI with the Genfity base_url — pattern is identical to OpenAI direct.

Setup

1Install: pip install langchain-openai.
2Instantiate ChatOpenAI with base_url + api_key.

https://python.langchain.com

LangChainpython

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="genfity/claude-opus-4.7",
    api_key="genfity_xxxxxxxxxxxx",
    base_url="https://ai.genfity.com/v1",
    temperature=0.2,
)

print(llm.invoke("Hello Genfity").content)

Vercel AI SDK

TypeScript toolkit for AI apps

Use @ai-sdk/openai with the createOpenAI({ baseURL }) helper.

Setup

1Install: npm i ai @ai-sdk/openai.
2Call createOpenAI with the Genfity baseURL.

https://sdk.vercel.ai

Vercel AI SDKtypescript

import { createOpenAI } from "@ai-sdk/openai";
import { generateText } from "ai";

const genfity = createOpenAI({
  apiKey: process.env.GENFITY_KEY!,
  baseURL: "https://ai.genfity.com/v1",
});

const { text } = await generateText({
  model: genfity("genfity/claude-opus-4.7"),
  prompt: "Hello Genfity",
});
console.log(text);

Laravel (PHP)

PHP HTTP client integration

Use Http::withToken to send requests to Genfity AI Gateway from Laravel.

Setup

1Set GENFITY_KEY in .env.
2Use Illuminate\Support\Facades\Http for POST requests.
3Parse the response with ->json().

https://laravel.com

Laravel (PHP)php

use Illuminate\Support\Facades\Http;

$response = Http::withToken(env('GENFITY_KEY'))
    ->post('https://ai.genfity.com/v1/chat/completions', [
        'model' => 'genfity/claude-opus-4.7',
        'messages' => [
            ['role' => 'user', 'content' => 'Hello Genfity'],
        ],
    ]);

return $response->json();

API Reference

Quick start

Authentication

Endpoints

Create Chat Completion (OpenAI)

Headers

Request Body

Anthropic Messages

Headers

Request Body

Count Tokens

Headers

Request Body

Embeddings

Headers

Request Body

Image Generation

Headers

Request Body

List Models

Headers

Model & Pricing

Error Codes

Rate Limits

IDE & Editor

Cline

Roo Code

Continue.dev

Cursor

Windsurf (Codeium)

Zed

Kilo Code

TRAE

Xcode

CLI & Terminal

Claude Code

opencode

Codex CLI

Aider

LLM (Simon Willison)

Goose

Crush

Droid (Factory)

Qwen Code

Grok CLI

OpenHands

OpenClaw

Desktop & Web UI

LibreChat

Open WebUI

Chatbox

Jan

LM Studio

Cherry Studio

SDK & Framework

OpenAI Python SDK

OpenAI Node SDK

Anthropic Python SDK

Anthropic Node SDK

LangChain

Vercel AI SDK

Laravel (PHP)

Schedule a Project Discussion Now!

Schedule a Project Discussion Now!

Schedule a Project Discussion Now!