| Model | Score | Status | Released | Price in/out | Context | |
|---|---|---|---|---|---|---|
| Qwen3.8-Flash lowest-price 1M-context coding and office agents | 8.4 | GA | 2026-08-26 | $0.15 / $0.47 | 1M | Review → |
| Qwen3.8-27B strongest Apache-2.0 model that fits one GPU | 8.5 | GA | 2026-08-14 | $0.50 / $3 | 1M | Review → |
| Qwen3.8-Max open-weight frontier agent for long-horizon coding and office work | 8.6 | GA | 2026-08-03 | $2 / $6 | 1M | Review → |
Qwen3.8-Max is Alibaba Cloud's flagship model, announced on 2026-08-03 as "the most capable model in the Qwen family to date" and the first Qwen Max-class model with downloadable weights. It is a 2.4-trillion-parameter sparse mixture-of-experts model (95B active) with a 1M-token context window, 131,072-token output, text, image and video input, and tunable reasoning depth. Alibaba built it for multi-day autonomous coding and professional office work, and now recommends it as its "highest capability" model in place of Qwen3.7-Max. The sentence a buyer needs: Qwen3.8-Max is a frontier-class agent model at $2 / $6 per 1M tokens with open weights under a custom license, independently ranked second on the LMArena vision board and sixth on its web-development board, but it is slow (about 39 tokens/sec) and most of its launch numbers come from Alibaba's own testing.
Full review →Qwen3.8-Flash is Alibaba Cloud's low-cost Qwen model, released August 26, 2026 on Model Studio and Qwen Cloud. It is the production version of the open-weight Qwen3.8-Flash-Next, a 125B-parameter mixture-of-experts model with 6B parameters active per token, which Alibaba describes as an early preview of the Qwen4 architecture. It reads text, images and video, holds 1M tokens of context, thinks by default at an adjustable effort level, and lists at $0.15 input / $0.47 output per 1M tokens, flat across the whole window. Artificial Analysis scores it 40 on its Intelligence Index, level with DeepSeek V4.1-Flash and ahead of GPT-5.6 Luna even at Luna's max effort, at the lowest list price in that group. For a buyer: it is the cheapest per-token route to near-flagship coding and office-agent quality. The costs are slow generation (about 57 tokens/sec), heavy token use, and a weights license that requires a separate deal for anyone running a model-hosting or coding/office-assistant business.
Full review →