Anthropic's official terminal coding agent — the gold standard, but $200-1000/month if you let it loose on a real project.
Claude Code is built for autonomous work — it reads your whole repo, plans long sessions, runs subagents in parallel, and re-sends growing context. A serious all-day session can hit 5M+ tokens. Anthropic Direct: ~$50-100/day. Through us: ~$10-20/day.
Sign up (free, no card), then generate an sk-cc-… key from your dashboard.
Set environment variables before launching claude.
export ANTHROPIC_BASE_URL=https://claudeapi.cheap/api/proxy
export ANTHROPIC_AUTH_TOKEN=sk-cc-your-key-here
# Then launch as usual:
claudeClaude Code keeps every feature. Only the API endpoint changed — your prompts, tools, and behaviour are unchanged.
Hard architectural work, planning
Default workhorse — Claude Code's recommended
Subagent fan-out, cheap parallel scans
Yes. Claude Code respects ANTHROPIC_BASE_URL exactly like the Anthropic SDK. Set it to https://claudeapi.cheap/api/proxy and use ANTHROPIC_AUTH_TOKEN with your sk-cc-... key.
Yes. Subagents are local orchestration in Claude Code; we just see the underlying API calls. All tools (Bash, Read, Edit, Write, WebFetch, etc.) work normally.
MCP servers run client-side. They never touch the LLM provider's API directly — Claude Code mediates. Our proxy is invisible to MCP.
No. Prompt caching is forwarded — you get cache writes (1.25× input) and cache reads (0.1× input) at our 80%-off price. Cache hit rate stays the same.
Free Basic plan, $19 lifetime Pro. Crypto only. No subscription.
Get an API key — free