Gemini 4 Argon rolls out, topping GPT-6 and Claude on most benchmarks
rickasaurus · x · 2026-10-14
Google's Gemini 4 Argon is rolling out, with an official eval table showing it leading GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5 on most benchmarks: Vals Index 68.9%, AutomationBench 51.3%, Harvey's Legal Agent 19.6%, DeepSWE 77.9%, Vibe Code Bench 91.9%, LABBench 2 88.8%, GraphWalks 1M-context 84.2%, Agent's Last Exam 39.5%, and LVBench 91.7%. It trails only on FrontierSWE, Terminal-bench 4.0, PostTrainBench, and OSWorld-2.0.
More from Models
- Gemini triggers SAFETY refusal mid-reverse-engineering, but finishes the dump first — banteg · 2026-10-14
- Gemini 4 Argon supports up to 1M token outputs, 8x any frontier GPT or Claude model — daniel_mac8 · 2026-10-14
- Brundage spots COT-like elements leaking into Argon's non-COT replies — Miles_Brundage · 2026-10-14
- NVIDIA open-sources PersonaPlex-7B, a full-duplex voice model that listens and speaks simultaneously — solyarisoftware · 2026-10-14
- Muse Wonder Tops TaxCalcBench at 71% — and Is First Model Caught Cheating — acbuller · 2026-10-14
- 'Two-way RL' for task-level model compression: building wafer-thin hyper-specialized models — tokenbender · 2026-10-14