Qwen3.8-Flash
Alibaba · Released 26 Aug 2026 · qwen3.8-flash
$0.15
Input price / 1M tokens
#3 of 61, cheapest first
$0.47
Output price / 1M tokens
#1 of 61, cheapest first
1M
Context window
#29 of 71
–
Output tokens / second
62.5%
SWE-bench Pro
#5 of 16
58.7%
DeepSWE v1.1
#16 of 17
91.7%
GPQA Diamond
#12 of 27
35.9%
Humanity’s Last Exam (no tools)
#13 of 23
Summary
Qwen3.8-Flash is an open-weight model from Alibaba, released on 26 Aug 2026.
At $0.15 input and $0.47 output per million tokens, it is #3 of 61 on input price, cheapest first.
Its 1M-token context window ranks #29 of 71.
Fast multimodal Qwen model based on the open-weight Qwen3.8-Flash-Next (125B total, 6B active, plus 51B n-gram embedding), an experimental preview of the Qwen4 architecture. The Qwen Community License allows commercial use with a name-display rule above 100M MAU or $20M monthly revenue, but any Model-as-a-Service or AI coding/office assistant business needs a separate license from Qwen regardless of size.
Benchmarks · 5 reported
| Benchmark | Score | Bar | Rank | Source |
|---|---|---|---|---|
| SWE-bench ProQwen3.8-Flash-Next open weights | 62.5% | #5 of 16 | Qwen3.8-Flash-Next model cardSelf-reported | |
| DeepSWE v1.1Qwen3.8-Flash-Next open weights | 58.7% | #16 of 17 | Qwen3.8-Flash-Next model cardSelf-reported | |
| GPQA DiamondQwen3.8-Flash-Next open weights | 91.7% | #12 of 27 | Qwen3.8-Flash-Next model cardSelf-reported | |
| Humanity’s Last Exam (no tools)no tools, Qwen3.8-Flash-Next open weights | 35.9% | #13 of 23 | Qwen3.8-Flash-Next model cardSelf-reported | |
| Toolathlon VerifiedPass@1, Qwen3.8-Flash-Next open weights | 73.5% | – | Qwen3.8-Flash-Next model cardSelf-reported |
Self-reported means the lab ran the test itself. A rank appears only where several labs report the same version of a benchmark.
Similar models
Models with the most benchmarks in common with Qwen3.8-Flash, and the closest scores.
SWE-bench Pro
#5 of 16Harder multi-file engineering tasks
- Qwen3.8-Flash62.5
- GLM-5.262.1
- Qwen3.7-Max60.6
- Qwen3.8-27B61.7
- GPT-5.6 Luna62.7
- GPT-5.558.6
GPQA Diamond
#12 of 27Graduate-level science questions
- Qwen3.8-Flash91.7
- GLM-5.291.2
- Qwen3.7-Max92.4
- Qwen3.8-27B89.2
- GPT-5.6 Luna92.3
- GPT-5.593.6
Humanity’s Last Exam (no tools)
#13 of 23Expert questions, no search or code
- Qwen3.8-Flash35.9
- GLM-5.240.5
- Qwen3.7-Max41.4
- Qwen3.8-27B30.8
- GPT-5.541.4
Compare Qwen3.8-Flash side by side
Pre-filled with the two closest models. Swap any of them.
Sources
- price: alibabacloud.com/help/en/model-studio/model-pricing, checked 2 Oct 2026
- context, max output, cache prices, modalities: qwencloud.com/models/qwen3.8-flash, checked 2 Oct 2026
- release date: alibabacloud.com/help/en/model-studio/newly-released-models, checked 2 Oct 2026
- weights, parameters, license, benchmarks: huggingface.co/Qwen/Qwen3.8-Flash-Next, checked 2 Oct 2026
Building on Qwen3.8-Flash?
Get the architecture right before the bill arrives.
We size caching, routing and fallbacks for your workload on a free call, and tell you where a cheaper model is good enough.