Qwen 3.8 27B launches as Apache 2 vision-capable LLM
Alibaba's Qwen lab released Qwen 3.8 27B, an Apache 2 licensed 27B parameter vision-capable LLM, with benchmarks showing boosts over Qwen 3.6 27B and Qwen 3.7-Plus. It defaults to xhigh reasoning effort, causing overthinking; Simon Willison recommends raising context to 262,144 tokens.
How this story unfolded
3 weeks · 27 reports · 151 community posts · 178 of 191 shown
- Aug 3
- Aug 4
- Aug 11
- Aug 12
- Aug 13
- Aug 14
- Aug 15
- Aug 16
- Aug 17
- Aug 18
Qwen 3.8 27B scores 52 on the Artificial Analysis Intelligence Indexsimonwillison.net
Qwen3.8-27B runs frontier-class coding agents and reasoning locally, no cloud API requiredventurebeat.com
huihui-ai/Huihui-Qwen3.8-27B-abliteratedhuggingface.co
HauhauCS/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUFhuggingface.co
orcarouter/Qwen3.8-27B-Uncensored-MLXhuggingface.co
huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUFhuggingface.co
orcarouter/Qwen3.8-27B-Uncensored-GGUFhuggingface.co
- Aug 19
- Aug 20
- Aug 21
- Aug 22
- Aug 23
Qwen by email
Get an email when Qwen has news
No news that day, no email.
More stories today
- Perplexity adds granular effort selector to Computer
- Mystery AI model Ox Alpha draws developers with free access
- GMKtec to launch Ryzen AI Max+ PRO 495 hardware at IFA Berlin 2026
- Drew Breunig: Fable's high cost ends the free lunch for coding models
- Grok voice mode adds two new voices: Aurora and Liora