Gemini 4 Scores Just 78% on Saturated FrontierMath Tier 4, Well Behind OpenAI's Models
haider1 · x · 2026-10-19
FrontierMath Tier 4 has been essentially saturated by OpenAI models—the chart shows GPT-6.1 Sol (max) at 100% and GPT-6 Astra variants at 97.6%, with Claude Opus 5.5 at 95%—while Google DeepMind's Gemini 4 Argon (high) managed only 78.0%±6.5%, level with GPT-5.5 Pro. The poster questions where Google's Deep Think is, noting DeepMind made big progress on research-level math months ago but hasn't shown it on this benchmark; its AI co-mathematician scored 75.6%.
More from Models
- Source revealed: Claude models' random number picks cluster on 42 and 47 — randal_olson · 2026-10-19
- 221 fresh chats: Claude Sonnet 5.5 says "47" for 92% of random number requests — randal_olson · 2026-10-19
- Bittensor subnet Ingot beats Qwen's own FP8 quantization within 4 days — const_reborn · 2026-10-19
- Cohere merges llama.cpp support, all its latest model architectures now compatible — nick_frosst · 2026-10-19
- Liquid AI explains designing LFMs for phones: less compute, better architecture and data — JosephJacks_ · 2026-10-19
- Report: Fable 5.5, Gemini 4 Carbon and GPT-6.1 Terra all due this month — Angaisb_ · 2026-10-19