LLM Providers
| Provider | Models | Free Tier | Get API Key |
|---|---|---|---|
| Google Gemini (recommended) | Gemini 3 Flash, 3.1 Pro, 3.1 Flash Lite, 2.5 Pro, 2.0 Flash, Gemma 4 | Yes | aistudio.google.com |
| OpenAI | GPT-5.4, GPT-5.4 Mini, GPT-5.4 Nano | No | platform.openai.com |
| Anthropic | Claude Opus 4.6, Sonnet 4.6, Haiku 4.5 | No | console.anthropic.com |
| OpenRouter | 300+ models (Claude, GPT, Llama, DeepSeek, etc.) | Yes (free models available) | openrouter.ai/keys |
| DeepSeek | DeepSeek V3 (chat), DeepSeek R1 (reasoning) | No | platform.deepseek.com |
| Groq | Llama 3.3 70B, Gemma 2 9B, Llama 3.1 8B | Yes (generous free tier) | console.groq.com |
| Mistral AI | Mistral Large, Medium, Small, Codestral | No | console.mistral.ai |
| Ollama (local) | Any model you pull locally | Free | ollama.com |
Switch provider or model at any time:
axiom-wiki modelGoogle Gemini is recommended because it has a generous free tier and strong performance for wiki tasks.
OpenRouter
Section titled “OpenRouter”OpenRouter is a unified API that gives you access to 300+ models from every major provider (Anthropic, OpenAI, Google, Meta, DeepSeek, Mistral, and more) through a single API key.
Use it when you want to:
- Access models from multiple providers without managing separate API keys
- Use free models (Llama 3.3 70B, DeepSeek R1, and others)
- Use the Auto router that picks the best model for each prompt automatically
- Try new models without switching providers
The setup is the same as any other provider — just paste your OpenRouter API key during axiom-wiki init.
DeepSeek
Section titled “DeepSeek”DeepSeek offers two models: V3 for general-purpose tasks (fast, very affordable) and R1 for chain-of-thought reasoning. Both use the OpenAI-compatible API format.
Groq provides extremely fast inference on open-source models. Great for interactive query mode where speed matters. Has a generous free tier.
Mistral AI
Section titled “Mistral AI”Mistral offers a range of models from the fast and cheap Small to the powerful Large. Codestral is a code-specialized model with a 256K context window — excellent for autowiki on large codebases.
Ollama
Section titled “Ollama”Ollama runs models locally with no API key. During setup, Axiom detects your installed models and lets you pick one. If no models are pulled, you can select a suggested model and Axiom pulls it automatically.
The default context window for Ollama is 65,536 tokens. Override it with ollamaNumCtx in your config. See the Ollama guide for details.