Skip to content

ollama-provider.md

Latest commit

 

History

History
71 lines (53 loc) · 2.31 KB

ollama-provider.md

File metadata and controls

71 lines (53 loc) · 2.31 KB

Ollama Native Provider

Claude Code Best supports Ollama through the native Ollama API, not the OpenAI-compatible endpoint. This lets Cloud and local Ollama use the same request shape for chat, tool calling, thinking, model discovery, and web utilities.

Configure

Use /login and choose Ollama, or configure these environment variables:

CLAUDE_CODE_USE_OLLAMA=1
OLLAMA_API_KEY=ollama_api_key
OLLAMA_BASE_URL=https://ollama.com/api
OLLAMA_DEFAULT_HAIKU_MODEL=qwen3:cloud
OLLAMA_DEFAULT_SONNET_MODEL=qwen3-coder
OLLAMA_DEFAULT_OPUS_MODEL=glm-4.7:cloud

OLLAMA_API_KEY is required for direct Ollama Cloud API access. It is not required for local Ollama. For local Ollama, set:

OLLAMA_BASE_URL=http://localhost:11434/api

If OLLAMA_BASE_URL is omitted, Claude Code Best uses https://ollama.com/api.

Model Mapping

Ollama model routing uses the same three Claude model families shown by /model:

  • OLLAMA_DEFAULT_HAIKU_MODEL
  • OLLAMA_DEFAULT_SONNET_MODEL
  • OLLAMA_DEFAULT_OPUS_MODEL

There is no global OLLAMA_MODEL override. This keeps Ollama behavior aligned with other third-party providers, where Haiku/Sonnet/Opus can map to different backend models.

When a direct Ollama model name is selected from /model or --model, it is sent to Ollama unchanged. When a Claude family model is selected, Claude Code Best maps it through the matching OLLAMA_DEFAULT_*_MODEL variable. If no family mapping is configured, the fallback is qwen3-coder.

Supported Features

  • Native POST /api/chat streaming
  • Ollama tool calling through tools
  • Ollama thinking through think
  • Native POST /api/web_search
  • Native POST /api/web_fetch
  • Dynamic context length discovery through POST /api/show
  • Local Ollama and Ollama Cloud through the same provider

The provider reads model context length from model_info.*.context_length or the num_ctx parameter returned by /api/show, then uses that value to choose the request output limit.

Known Limits

Ollama does not expose an official Cloud quota or remaining-balance API in the documented native API. Claude Code Best therefore does not show Ollama Cloud remaining quota.

Anthropic-only server tools are not sent directly to Ollama. Web search and web fetch are handled client-side through Ollama's native web APIs when the Ollama provider is active.