Home / Guides

Free LLM API guides

Step-by-step guides to free LLM APIs: rate limits, setup for coding tools, capabilities, errors and comparisons. Each guide uses live data from the FreeLLMAPI catalog (87 providers, checked 19 November 2026).

Free API data

  • Do free LLM APIs train on your data?: Which free LLM API tiers may use your prompts for training or human review, from each provider's terms. Gemini, Groq, NVIDIA, Cohere, OpenRouter and more.
  • Free LLM API for commercial use: Which free LLM API tiers allow commercial or production use, from provider terms. Cohere trial keys and NVIDIA trial access do not. Gemini, Groq and others.
  • Free LLM API rate limits: Free tier rate limits for 45 LLM providers in one live table: requests per minute, per day and tokens per minute, with how to read and stack them.
  • Free LLM API with no credit card: Which LLM providers give a free API key without a credit card, checked on their own pages. Signup facts, free models and the limits of each free tier.
  • Free LLM APIs hosted in Europe: Free or low-cost LLM API tiers hosted in the EU: OVHcloud, Scaleway and Mistral. Where each one processes data, training rules and Gemini rules for EU users.
  • Free OpenAI-compatible API base URLs: The verified OpenAI-compatible base URL of each free LLM provider, with auth notes and one example. Or use one local base URL for all free providers.
  • LLM API without signup: LLM APIs you can call with no account and no key: OVHcloud, Kilo Gateway and AI Horde. Exact anonymous limits, privacy notes and honest caveats.

Coding tools and apps

  • Cline, Roo Code and Kilo Code with a free API: Run Cline, Kilo Code or Roo Code on free LLM models. One setup command or the OpenAI Compatible provider fields, plus tool-calling models that work for agents.
  • Continue with a free LLM API: Set up Continue in VS Code or JetBrains with free LLM models: config.yaml for chat and agent mode, plus tab autocomplete on /v1/completions. Exact config.
  • Free AI models in VS Code: Use free AI models in VS Code: Copilot bring your own key with a Custom Endpoint, Cline or Kilo Code for agents, and Continue for free ghost-text autocomplete.
  • Free API for SillyTavern: Use free LLM tiers in SillyTavern with the Custom (OpenAI-compatible) source. Set the endpoint, select large-context models and keep long chats in limits.
  • Free LLM API for n8n: Run n8n AI workflows on free LLM tiers. Set the Base URL in the OpenAI credential, use the AI Agent node with tool calling, and retry when a free limit is hit.
  • Free LLM API for Open WebUI: Connect Open WebUI to free LLM tiers with one OpenAI-compatible connection. Set the URL and key, keep the model list short, and fail over at each limit.
  • Gemini CLI with a custom endpoint and free models: Point Gemini CLI at a custom endpoint with GOOGLE_GEMINI_BASE_URL and run it on free models. Setup, family mapping, the 2026 free tier change and real errors.
  • How to use Claude Code for free: Run Claude Code on free LLM tiers. Set two environment variables, pick models that handle long agent prompts, and fail over when one provider hits its limit.
  • Use Aider with a free LLM API: Run Aider on free LLM models. The exact .aider.conf.yml, the openai/ model prefix, edit formats for free models, weak model settings and real Aider warnings.
  • Use Codex CLI with free models: Run OpenAI Codex CLI on free LLM models from other providers. The exact config.toml provider block on the Responses API, the env key, model choice and fixes.
  • Use Cursor with a free LLM API: Connect Cursor chat to free LLM models with a custom OpenAI key and base URL. What works, which plans allow own keys, the HTTPS step and which models to use.
  • Use free AI models in Zed: Connect the Zed agent panel to free LLM models with an OpenAI-compatible provider or Ollama emulation. Exact settings.json, key setup, JetBrains and fixes.
  • Use OpenCode with a free LLM API: Run the OpenCode coding agent on free LLM models. One setup command, the exact opencode.json provider block, where the key goes and which free models to select.

Frameworks and code

  • Free AI API for a Discord or Telegram bot: Build a Discord or Telegram AI bot on free LLM tiers. Complete Python code with chat memory, per-user limits, message splitting and 429 handling.
  • Free LLM API for AI agents: Run AI agents on free LLM tiers. Use tool-calling models, an agent loop with retries, routing strategies per step, and the gateway's own MCP server.
  • Free LLM API for LangChain: Use free LLM tiers in LangChain with ChatOpenAI and a custom base_url. Python and JavaScript code for chat, tools, agents, embeddings and 429 retries.
  • Free LLM API in Python: Call free LLM tiers from Python with the official openai package. Complete code for chat, streaming, tools, async batches, embeddings and 429 retries.
  • Free LLM for LlamaIndex: Run LlamaIndex on free LLM tiers. Set OpenAILike and OpenAILikeEmbedding to one free gateway, build a RAG index and keep ingestion inside free rate limits.

Capabilities

  • Best free coding model APIs: The best free coding LLM APIs today: agent models with tool calling and long context, plus fast autocomplete models. Live limits from the FreeLLMAPI catalog.
  • Free embeddings API: Free text embeddings through one OpenAI-compatible endpoint. Compare dimensions, input limits and free quotas, then call /v1/embeddings in curl or Python.
  • Free image generation API: Generate images free with FLUX on Cloudflare Workers AI and other providers through one /v1/images/generations endpoint. Real limits, code and fixes.
  • Free LLM API for RAG: Build retrieval-augmented generation on free LLM tiers. Free embedding models, long-context chat models, a rerank step and complete Python code with limits.
  • Free LLM API with 1 million token context: Free LLMs with 1M-token context windows from a live table. Learn the tokens-per-minute limits that stop large prompts, and how to count tokens first.
  • Free LLM API with function calling: Free LLMs that support OpenAI-style function calling, from a live table. Working tool loop in curl and Python, streaming, and the common pitfalls.
  • Free reasoning model API: Free reasoning models such as GPT-OSS, DeepSeek R1 Distill, QwQ and Qwen Thinking from a live table. Thinking token costs, empty answers and code.
  • Free speech to text API: Transcribe audio free with Whisper, Deepgram Nova-3, Voxtral and more behind one /v1/audio/transcriptions endpoint. Limits, formats and code.
  • Free text to speech API: Free text to speech from Groq, Gemini, Cloudflare, Speechify and Cartesia behind one /v1/audio/speech endpoint. Character limits, formats and code.
  • Free translation API with LLMs: Translate text with free LLM tiers through one OpenAI-compatible key. Multilingual free models, batch code that keeps placeholders, and the limits to plan for.
  • Free vision API: Send images to free vision LLMs such as Gemini, Qwen and Gemma through one OpenAI-compatible endpoint. Live model table, size limits and code.
  • Free web search API for AI agents: Give an LLM live web results free. Compare Tavily, Exa, Firecrawl and Linkup free credits, call /v1/search, and use Gemini Google Search grounding.

Rate limits and errors

  • Gemini API quota exceeded (429 RESOURCE_EXHAUSTED): Fix the Gemini API 429 RESOURCE_EXHAUSTED and quota exceeded errors on the free tier. Exact limits, what each message means, and how to fail over.
  • Groq rate limit reached (429): Fix Groq 'Rate limit reached' 429 and 413 errors on the free tier. Limits per model, the rate limit headers to read, retry code and failover.
  • NVIDIA NIM API rate limit: The NVIDIA NIM free API rate limit explained: the common 40 RPM cap, why requests to raise it fail, what the trial terms allow, and how to spread the load.

Providers

Models

  • Free MiniMax API: MiniMax has no free text API tier, but MiniMax M3, M2.7 and M2.5 run free on several third-party tiers. See the live free routes, limits and setup steps.
  • Free Whisper API: Use OpenAI Whisper free on Groq, Cloudflare Workers AI and OVH through one /v1/audio/transcriptions endpoint. File limits, model choice and code.
  • Is there a free Claude API?: Anthropic has no free Claude API tier. See the legitimate options: plan credits, trial routes and a free Claude-format endpoint that runs other free models.
  • Is there a free GPT API?: OpenAI has no free API tier with free tokens. See the honest options: GPT-OSS open-weight models on free tiers, data sharing tokens and GPT routes on gateways.

Comparisons and alternatives

  • Groq vs Cerebras: Groq vs Cerebras in 2026: output speed, free plan vs trial credit, rate limits and models side by side, with a clear choice for each use case.
  • Helicone alternative: Helicone joined Mintlify in March 2026 and runs in maintenance mode. Compare alternatives for LLM observability and routing, with free self-hosted options.
  • LiteLLM alternative: Looking for a LiteLLM alternative? Compare LiteLLM with a self-hosted gateway that pools free LLM tiers, plus OpenRouter and Portkey, with a fair table.

Browse all free models