A Local 27B Model Is Now 35% Smarter Than Last Year's SOTA
cephaloform · x · 2026-10-15
- A viral post cites the Artificial Analysis Intelligence Index to argue that today's locally runnable 27B models (e.g. Qwen 3.8 27B) are roughly 35% more intelligent than the leading SOTA models from just one year ago.
- The chart places Qwen3.8 27B 272B (estimated score 34) at the top, ahead of GPT-5, Claude 4.1 Opus, and Grok 4.
- Takeaway: frontier-level capability went from cloud-only to locally runnable in about a year — prompting the question of what will run locally next year. Note the Qwen3.8 score is an estimate pending independent evaluation.
More from Models
- Runway CEO reacts to claim it brought DeepMind's world model theory to life first — c_valenzuelab · 2026-10-15
- Argon 'Nails' Data Analysis and Excel Tasks in Production, Claims Dev: Better Than Opus and Cheaper — bindureddy · 2026-10-15
- Gemini 4 Argon scores 78% on FrontierMath T4; Qwen3.8 Max still tops Chinese models at 46% — teortaxesTex · 2026-10-15
- Pointing-at-things eval: Gemini 4 Argon leads, Mistral Large 4's SOTA vision claim falls flat — pbaylies · 2026-10-15
- Developer says OpenAI deleted his account for the fifth time, appeal auto-rejected — QuixiAI · 2026-10-15
- LFM2.5-24B-A2B hits ~300 tok/s on a single DGX Spark via TandemLLM — JosephJacks_ · 2026-10-15