LaneZero
Docs Benchmarks GitHub Sign in
The fast lane
to every LLM.
Any model, one API, your keys. Rust-built, microsecond overhead.
22% faster than Vercel AI Gateway 8% faster than OpenRouter see the benchmarks
Sign in with GitHub Read the docs
$ curl https://api.lanezero.dev/v1/chat/completions \
-H "Authorization: Bearer rw-live_..." \
-d '{ "model": "provider/model-id", "stream": true, ... }'
12 providers. One API.
341 models and counting. Bring a key for any of them.
Anthropic
OpenAI
Google
xAI
DeepSeek
Qwen
Moonshot
Z.ai
MiniMax
Tencent
Mistral
Meta
BYOK, in a real vault
Your keys are envelope-encrypted and wiped from memory after use. Prefer passthrough? Send a key per request and we store nothing.
Built to be benchmarked
LaneZero is Rust with nothing slow on the hot path. Every number we publish is reproducible. Run our open harness and check us.
Fallbacks on the grid
If a provider rate-limits or fails, we retry the next one in your list. Your request still finishes.
Your keys, your providers' bill, and nothing in between. No credits, no markups, no billing.
DocsGitHubStatusTermsPrivacy