Models

Models

Open-weight models we serve on our own GPUs, through OpenAI- and Anthropic-compatible APIs. The numbers are measured on live traffic, not on a benchmark.

Updated · Live figures load from api.tokenflow.dev

Live: last 24 hours. Context is prompt plus output, in tokens.
ModelStatusFirst tokenOutputSuccessContextInput
qwen3.8-27bGeneral chat, coding agents and tool use. Our default model. — — — — 128K Text, Images
qwen3.8-27b-directQwen3.8 27B with an adapter that declines fewer requests. — — — — 128K Text, Images
qwen3.8-27b-cuteThe direct model with a casual, short, emoji-friendly chat persona. — — — — 128K Text, Images
gemma4-12bSmaller and faster. Good for short chat, extraction and image questions. — — — — 128K Text, Images
gemma4-12b-directGemma 4 12B with an adapter that declines fewer requests. — — — — 128K Text, Images

Live figures cover all API traffic in the last 24 hours and refresh every minute. First token is the median time from request to the first streamed token, at any prompt length. Output is output tokens per second after the first token, over answers of 16 tokens or more. Success is the share of requests that did not fail on our side; client errors such as a bad request or an exhausted quota do not count against it. A figure needs at least 20 requests behind it; until then it shows a dash.

GET /v1/models with your API key lists the models your key can use. The same live figures, without a key: GET https://api.tokenflow.dev/v1/public/models.