Added a faster lane for the model most of you use for free coding and agent work. The new id is qwen3.8-27b-1:free. It runs the same open-weight Qwen3.8 27B, served from a new provider, and it allows about 1,000 requests a day instead of the 100 the existing lane allows. The original qwen3.8-27b:free stays where it is, on the same 100-a-day terms as before.
Why a second id instead of speeding up the first one: the two lanes come from different upstream providers, each with its own daily cap. Giving the faster lane its own id means you can pick it on purpose, and a burst of traffic on one id no longer eats into the other. Both ids are live now, and both appear in the model catalog and the free lane.
What you get
It is the same model, so the change is about throughput, not quality:
- About 1,000 requests a day on
qwen3.8-27b-1:free, against 100 onqwen3.8-27b:free. Measured on the provider’s own limits, per model. - 131K context window on the new lane. The older lane advertises more (204K), so if you need a very long prompt, keep using
qwen3.8-27b:free. - Text in, text out, tools and streaming, on the ordinary chat completions endpoint. It bills $0 per token: free-lane pricing, no card, counted against the free-lane daily allowance rather than your credit.
How to use it
Point your existing client at the gateway and set the model:
curl https://api.oneaiapikey.com/v1/chat/completions \
-H "Authorization: Bearer $ONEAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen3.8-27b-1:free",
"messages": [{"role": "user", "content": "Explain what a gateway does in one line."}]
}'If a script already pins qwen3.8-27b:free, you do not have to change anything. It keeps working. Switching to the new id is a one-line change and buys you the higher daily cap. Both ids sit next to the other free models — see StepAudio 3 Chat Preview, the newest free addition, and the full free models list.
Verified before it went live
We do not list a model until it answers. Checked 2026-11-13: the new id appears in GET /v1/models, and a real request through the gateway returned a completion, billed $0. The upstream answers quickly on short prompts. For live health, watch the status board.
One honest note on limits: the 1,000-a-day cap is on the upstream provider and is shared by everyone calling the new id, not just you. When it is spent for the day, requests on this id fall back to the existing lane and then to the models that always answer, the same way every free id here behaves. The rate limits docs cover how the allowance refills at 00:00 UTC.
Watch the changelog for the next additions.