Oct 9, 2026

New chat models: Qwen3.8 Max and Qwen3.8 Flash are serving now

Two models from Alibaba Cloud Model Studio are on POST /v1/chat/completions: qwen3.8-max for long context and harder requests, and qwen3.8-flash for summaries, classification and anything you run in volume. Change the model and nothing else — streaming and tool calling work as they do for the other models. qwen3.8-max is 60 AP per 1K input tokens and 180 AP per 1K output; qwen3.8-flash is 5 and 15. Every price is in GET /v1/models — read it from there rather than keeping a copy. These models can think before they answer, and here that is off unless you ask for it. Upstream it is on by default, the thinking is billed as output, and it does not count against max_tokens: we sent max_tokens: 40 and got 3,216 tokens back after 67 seconds. So on api.aicu.ai, max_tokens (or max_completion_tokens) caps thinking and answer together, and the models answer directly. To have them think, send reasoning_effort (minimal, low, medium, high); the thinking then comes back in reasoning_content and is charged as output. Leave room in the cap when you do: a small cap can be spent entirely on thinking, and the answer comes back empty with finish_reason: "length".