Developers claim unreleased Gemini 4 Argon scores 91.9 atop VibeCodeBench
sunjiao123sun_ · x · 2026-10-01
Developer sunjiao123sun quote-tweets Google's Kuba saying "without Kuba, Gemini 4 Argon won't get 91.9 on VibeCodeBench." Kuba himself posted that he worked on vibe coding capabilities for Gemini 4 and credited the team for putting it at the top of Vibe Code Bench. This is a credible but unofficial leak about the unreleased Gemini 4 Argon model, not yet confirmed by Google.
Related event: Leaked Gemini 4 Argon Benchmarks Top Multiple Leaderboards(3 posts)→
More from Models
- First look at Opus 5.5 building an SCP-096 game in one go — imjustnewatai · 2026-10-01
- Singapore punches above its weight in AI as the world's No.4, but Claude's web integration lags OpenAI — teortaxesTex · 2026-10-01
- Hidden Dates in System Prompts Swing LLM Eval Scores by Up to 14% — Mario Sanz-Guerrero · 2026-10-01
- burny pushes back on 'LLMs just predict likely text': RLHF, CoT and tools changed that — burny_tech · 2026-10-01
- Leaked Astra CoTs Suggest OpenAI Models Are Trained With Hardcoded Thinking-Budget Prompts — tokenbender · 2026-10-01
- Chinese embodied AI model Maxwell tops global ranking on physical tasks — scmp_news · 2026-10-01