Claude Code makes local LLMs 90% slower

https://unsloth.ai/docs/basics/claude-code

by telotortium • 4 months ago

4 1 4 months ago

falcor84• 4 months ago

The relevant part is that Claude Code adds an Attribution header, which invalidates the KV Cache for local models. The fix proposed here is to set the following in ~/.claude/settings.json when using Claude Code against a local model:

    "env": {
      ...,
      "CLAUDE_CODE_ATTRIBUTION_HEADER" : "0"
    },

Claude Code makes local LLMs 90% slower

Related Stories