Qwen3.8-27B matches frontier models on agentic benchmarks
Qwen3.8-27B scores 52 on the Artificial Analysis Intelligence Index, matching GPT-5.6 Luna and DeepSeek V4. It runs on a single RTX 3090 at ~60 tokens/s, with a 256k context window.
How this story unfolded
7 days · 0 reports · 23 community posts · 23 of 24 shown
- Aug 14
- Aug 15
- Aug 16
- Aug 17
- Aug 18
- Aug 19
- Aug 20
- Aug 21
Qwen by email
Get an email when Qwen has news
No news that day, no email.
More stories today
- LLM beats Twilight Struggle bot in Cold War board game experiment
- OpenAI urges California to strengthen AI safety bill SB 53
- FTC targets AI-driven personalized pricing
- Simon Willison: key skill for coding agents is confident instruction and verification
- Anima-3.8B with Qwen-3.5 4B released by lylogummy