Models - Fortytwo

	Qwen3.8 27B
Description	Compact model for agentic coding and multi-step tasks
Model ID	`qwen/qwen3.8-27b`
Creator	Qwen team, Alibaba Cloud
Licence	Apache 2.0
Input	Text, image
Output	Text
Context window (tokens)	262,144
Max output (tokens)	235,929
Streaming	Yes
Reasoning effort	`none`, `minimal`, `low`, `medium` (default), `high`, `xhigh`, `max`
Tool calling	Yes, including parallel calls
Structured output	`json_object`, `json_schema` with `strict: true`
Sampling parameters	`temperature`, `top_p`, `top_k`, `min_p`, `presence_penalty`, `repetition_penalty`. Not `stop` or `seed`
Image input	PNG, JPEG, WebP, by HTTPS URL or base64 data URL
Endpoint	`POST /v1/chat/completions`

This table is the model catalogue referred to in the Terms of Service. Each model’s licence is linked in it. Send the model ID in the model field. Prices are on Pricing, rate limits on Rate limits.

Not supported

Video, audio and file input, even where the upstream model card lists them. Such a request fails with 400 unsupported_content_type.
Fields the API does not support, such as frequency_penalty or logprobs, fail with 400 invalid_request — see OpenAI compatibility.

List model IDs

GET /v1/models returns the IDs the API accepts now. Each entry has id, object, created and owned_by only.

curl https://api.fortytwo.network/v1/models \
  -H "Authorization: Bearer $FORTYTWO_API_KEY"

There is no endpoint for a single model: GET /v1/models/{id} returns 404.

Model retirement

A generally available model is retired only after at least 30 days’ notice where reasonably practicable; the exceptions are in Terms of Service, section 10.

Last modified on October 3, 2026