Mollick: Qwen 27B is a great local model but nowhere near rivals on agentic tasks
emollick · x · 2026-08-19
Wharton professor Ethan Mollick pushed back on the claim that Qwen 27B is open source's "DeepSeek moment": while it's a really good local model, when you actually use it, it's immediately and obviously far behind the other models on the leaderboard for agentic tasks — especially the complex tasks GDPval-AA purports to measure. His advice: do your own benchmarking rather than trusting the narrative.
More from Models
- Opus 5.5 value may end the habit of using the top model for everything — adonis_singh · 2026-10-16
- Peter Yang: Opus 5.5 Still Struggles to Hold Its Ground When Pushed Back — petergyang · 2026-10-16
- Robot Startup Ditches VLAs for Frontier LLMs in a Customer Deployment — ihorbeaver · 2026-10-16
- Musk amplifies user claim that Grok's computer use is fast enough to skip a Codex subscription — elonmusk · 2026-10-16
- PhoneLLM launches open voice-AI models with sub-100ms latency, tops PhoneBench at 83% — charles_irl · 2026-10-16
- Transluce vets: a year of neuron autointerp work still can't explain complex behaviors — aryaman2020 · 2026-10-16