Guide · LiteLLM
Connect LiteLLM to Synap
-
Get a free key
Sign in with Google at synap.linkrra.com — no card. Your key starts with
sk-pool-. -
Call it from Python
The
openai/prefix selects the OpenAI-compatible transport; what follows it is the Synap model id.# pip install litellm import os from litellm import completion r = completion( model="openai/qwen3-coder-30b", api_base="https://synap.linkrra.com/v1", api_key=os.environ["SYNAP_API_KEY"], messages=[{"role": "user", "content": "Write a haiku about diffs"}], ) print(r.choices[0].message.content) -
Or put it behind the proxy
One entry in
config.yaml, and every OpenAI client in your stack can reach Synap through the gateway. The proxy will not start without a master key of its own.# config.yaml model_list: - model_name: qwen3-coder-30b litellm_params: model: openai/qwen3-coder-30b api_base: https://synap.linkrra.com/v1 api_key: os.environ/SYNAP_API_KEY# pip install 'litellm[proxy]' export SYNAP_API_KEY=sk-pool-... export LITELLM_MASTER_KEY=sk-$(openssl rand -hex 32) litellm --config config.yaml --port 4000 curl http://127.0.0.1:4000/v1/chat/completions \ -H "authorization: Bearer $LITELLM_MASTER_KEY" \ -H 'content-type: application/json' \ -d '{"model":"qwen3-coder-30b","messages":[{"role":"user","content":"Say OK"}]}'
Notes, prices and limits
The openai/ prefix, as the SDK or behind the proxy. Recommended model: qwen3-coder-30b —
$0.0675 in and $0.27 out
per 1M tokens, 98,304-token context.
Run live against this API with litellm 1.106.6, both ways: the SDK call, and a real completion fetched through a locally started proxy.
litellm 1.106.6 refuses to start the proxy with no master key set — it exits and tells you to set LITELLM_MASTER_KEY (or general_settings.master_key). That is the key the proxy’s own callers present; your sk-pool- key stays in the config and is never handed to them.
Keeping model_name equal to the Synap model id means callers learn nothing new. A different name maps fine too — litellm_params.model is what reaches us.
On the free tier: every request counts input and output against 100,000 free tokens a day, and answers stop at 4,096 output tokens. Coding agents send long prompts, so the free tier covers a short session; credits cover the rest.
Field names from LiteLLM docs: OpenAI-compatible endpoints, run against this API, checked 2026-11-03 17:45 UTC. Prices refresh from /v1/models when this page loads.