Qwen3.8-2.4T-A95B is a Qwen model with 262K context, $2 in / $6 out per million tokens, last verified October 2, 2026. Open weights: yes. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.
Qwen's downloadable text-only Max-class checkpoint, with hybrid attention and tunable reasoning, distinct from the enhanced managed Qwen3.8-Max model.
Recorded profile verification October 2, 2026·Individual provider fields may change
Pricing tiers: Alibaba Cloud lists the hosted qwen3.8-2.4t-a95b endpoint at $2/$6 per MTok in Singapore International scope; Beijing is $1.65/$4.951. These are hosted regional fees, not self-hosting costs. Native checkpoint context is 262,144, extensible to 1,010,000 with documented serving configuration; the hosted pricing row uses up to 1M input, not a native configuration or output cap. Its pricing row lists both thinking modes, while the downloadable checkpoint card excludes non-thinking mode; do not transfer hosted gateway behavior to local weights.
Best for
Open-weight deployments
Text-only agents
Reasoning
Watch out
The August 13 QwenCloud announcement documents the hosted/open-version listing, not the exact weight-publication day. The checkpoint card’s benchmark column is Qwen3.8-Max, so those scores are not inherited here. This text-only checkpoint lacks managed Max’s vision input and non-thinking mode. Low/medium/xhigh default to xhigh; preserve_thinking defaults true. Recommended sampling: temperature=1.0/top_p=0.95/top_k=20. The exact custom licence includes attribution and separate commercial-licensing conditions; read it. A hard output cap remains unverified.
Source receipts
Catalog figures for this model were checked against the following sources.
Answered from the verified figures on this page rather than general guidance.
How much does Qwen3.8-2.4T-A95B cost per million tokens?
Qwen3.8-2.4T-A95B lists base rates of $2 per million input tokens and $6 per million output tokens. Estimate both input and output token volume and any additional charges for your workload. Alibaba Cloud lists the hosted qwen3.8-2.4t-a95b endpoint at $2/$6 per MTok in Singapore International scope; Beijing is $1.65/$4.951. These are hosted regional fees, not self-hosting costs. Native checkpoint context is 262,144, extensible to 1,010,000 with documented serving configuration; the hosted pricing row uses up to 1M input, not a native configuration or output cap. Its pricing row lists both thinking modes, while the downloadable checkpoint card excludes non-thinking mode; do not transfer hosted gateway behavior to local weights.
What is Qwen3.8-2.4T-A95B's context window?
Qwen3.8-2.4T-A95B accepts about 262K tokens of context. That only matters if you routinely send very long documents, large codebases, or multi-turn histories that approach that limit.
What is Qwen3.8-2.4T-A95B best for?
Qwen3.8-2.4T-A95B is a frontier tier from Qwen. It suits open-weight deployments, text-only agents, reasoning. The August 13 QwenCloud announcement documents the hosted/open-version listing, not the exact weight-publication day. The checkpoint card’s benchmark column is Qwen3.8-Max, so those scores are not inherited here. This text-only checkpoint lacks managed Max’s vision input and non-thinking mode. Low/medium/xhigh default to xhigh; preserve_thinking defaults true. Recommended sampling: temperature=1.0/top_p=0.95/top_k=20. The exact custom licence includes attribution and separate commercial-licensing conditions; read it. A hard output cap remains unverified.
Can I self-host Qwen3.8-2.4T-A95B?
Qwen3.8-2.4T-A95B publishes open weights, but self-hosting depends on the licence, hardware footprint, quantisation quality, and serving stack. A hosted API is often cheaper until you have measured throughput and concurrency on your own hardware.
