freeapis

Documentation

Integrations

Create one freeapis key, then connect it to Codex, Copilot CLI, or Claude Code. These are the first supported tools; more clients are in development.

Models & Limits

Our API key currently provides DeepSeek V4 Flash, MiMo V2.5, and Fusion MoA as free models, as well as GPT-5.5 / GPT-5.4 model variants. Free quota can be increased through Reward tasks, while GPT models use permanent quota.

Usage limits: free quota provides 5,000,000 tokens per 5 hours and 35,000,000 tokens per week. Deductions are calculated precisely based on model prices.

OpenAI-compatible endpointcurl https://freeapis.xyz/v1/models \ -H "Authorization: Bearer fapi_your_key"

CODEX

Codex integration

Codex reads provider settings from ~/.codex/config.toml. Add freeapis as a custom provider and point Codex at one of the supported free models.

# 1. Install and run the standalone Codex CLI.
# 2. Put this in your Codex config directory, e.g.
#    ~/.codex-freeapis/config.toml

model = "fusion"
model_provider = "freeapis"
model_reasoning_effort = "medium"

[model_providers.freeapis]
name = "freeapis"
base_url = "https://freeapis.xyz/v1"
wire_api = "responses"
env_key = "FREEAPIS_API_KEY"

# 3. Start Codex with the isolated config.
export CODEX_HOME="$HOME/.codex-freeapis"
export FREEAPIS_API_KEY="fapi_your_key"
codex

For GPT effort routing, keep model as the base model and set model_reasoning_effort, for example model = "gpt-5.5" and model_reasoning_effort = "high".

COPILOT CLI

Copilot CLI integration

Run the standalone Copilot CLI with a custom OpenAI-compatible Responses API provider. Use your freeapis key as the bearer token and point the provider at https://freeapis.xyz/v1.

# Install and run the standalone Copilot CLI.
# Use an isolated home so this test does not touch your normal Copilot state.
export COPILOT_HOME="$HOME/.copilot-freeapis"
export COPILOT_AUTO_UPDATE=false

export COPILOT_PROVIDER_TYPE=openai
export COPILOT_PROVIDER_BASE_URL="https://freeapis.xyz/v1"
export COPILOT_PROVIDER_BEARER_TOKEN="fapi_your_key"
export COPILOT_PROVIDER_WIRE_API=responses
export COPILOT_PROVIDER_TRANSPORT=http

export COPILOT_MODEL="fusion"
export COPILOT_PROVIDER_MODEL_ID="fusion"
export COPILOT_PROVIDER_WIRE_MODEL="fusion"

copilot

Use a base model such as fusion, deepseek-v4-flash, mimo-v2.5, gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna, gpt-5.5, gpt-5.4, or gpt-5.4-mini.

CLAUDE CODE

Claude Code integration

Run the standalone Claude Code CLI against the Anthropic-compatible freeapis endpoint. Use your freeapis key as ANTHROPIC_AUTH_TOKEN and pass the model explicitly.

# Install and run the standalone Claude Code CLI.
# Use an isolated config dir for this test.
export CLAUDE_CONFIG_DIR="$HOME/.claude-freeapis"
export ANTHROPIC_BASE_URL="https://freeapis.xyz/anthropic"
export ANTHROPIC_AUTH_TOKEN="fapi_your_key"

claude --model fusion

Direct Claude Code usage should pass --model explicitly. The Anthropic-compatible endpoint is https://freeapis.xyz/anthropic/v1/messages.

API key