LiveBench 2.0: Gemini 4 Argon Trails Claude Fable and GPT-6 Astra but Beats Open Source
bindureddy · x · 2026-10-14
Bindu Reddy unveiled the first results from LiveBench 2.0, a new benchmark he says is hard to benchmaxx and reflects real-world usage:
- Claude Fable 5.1 tops the board at 88.0 overall, but is the priciest at $4.526 per successful task
- Claude Opus 5.5 (86.5) and GPT-6 Astra (86.1) follow, with Astra at $3.442 per task
- GPT-6.1 Sol scores 85.4 at just $0.734 per task — the clear price-performance pick
- Gemini 4 Argon lands at 84.1, behind the closed-source pack despite the best reasoning score (95.1), but still ahead of open-source DeepSeek Flash and well above Muse Wonder
Breakdowns cover agentic, data analysis and design dimensions, with per-successful-task cost as a first-class metric.
More from Models
- Sonnet 5.5 landing on the free tier is more significant than people realize — adonis_singh · 2026-10-14
- Copilot keeps hitting company guardrails on tasks its chat mode handles fine — annetgriffin · 2026-10-14
- Gemini 4 Argon reportedly beats human baseline on all four stealth capability evals — scaling01 · 2026-10-14
- User praises Gemini thought summaries but wants full raw traces — QuixiAI · 2026-10-14
- Gemini 4 Argon lands in OpenCode as team's second most-used model — adonis_singh · 2026-10-14
- Claude Opus 5.5 vs Gemini Argon: Same 3D Island Prompt, Strikingly Different Results — adonis_singh · 2026-10-14