Gemini 3.1 Pro Still the GOAT in Most Benchmarks Except Coding
Last_Conclusion_8984 · reddit · 2026-08-20
A Reddit user claims that Gemini 3.1 Pro remains the best model (GOAT) in almost every benchmark, excluding coding and long-horizon tasks.
More from Models
- Stanford Framework Boosts DeepSeek Past Claude at 1/11th Cost — FuSheng_0306 · 2026-08-20
- Qwen3.8-Max tops frontend code leaderboard, beating Claude Fable 5 — Alibaba_Qwen · 2026-08-20
- Qwen3-Powered ASR Model superwhisper/s1-mini Trends on Hugging Face — superwhisper · 2026-08-20
- User tests reveal Qwen3.8-27B knowledge regression vs 3.6 — EmPips · 2026-08-20
- Qwen3.8 27B Speeds Up 3x Post-Release: Open Source Advantage — TheMoonMidas · 2026-08-20
- Mac can now run a 27B model locally that codes, reasons, and sees — TheMoonMidas · 2026-08-20