RWE-Bench: GPT-6 Astra leads real-world engineering eval at 37%, most models score 0%
wavefnx · x · 2026-10-28
Xplore launched the RWE-Bench leaderboard, scoring 11 frontier models on 20 real physical engineering tasks—engine component design, circuit board routing, turbine vane flow simulation—with physics solvers requiring every engineer-set requirement to pass. GPT-6 Astra leads at 37%, followed by Claude Opus 5.5 (29%), GPT-6.1 Sol (25%), Gemini 4 Argon (12%), Claude Fable 5.1 (7%), and Kimi K3.1 Pro (1%), while DeepSeek, GLM-5.3, Grok 4.7, Muse Wonder, and Qwen3.8 Max all scored 0%. Cost per run ranges from $1.03 (DeepSeek) to $97.57 (Claude Fable). Commenter wavefnx notes results are task-dependent: all models make basic architecture mistakes without precise steering.
More from Models
- OpenAI opens Decisions API beta: up to 10x faster decisions than GPT-6 Luna — OpenAIDevs · 2026-10-28
- Open-Weight AI Security Research Stalls as $80 Fine-Tuning Undoes Safety Interventions — Miles_Brundage · 2026-10-28
- Socher's Recursive Unveils Flint: 125+ tok/s Local Coding Model Built by Automated Research — jeffclune · 2026-10-28
- Liquid AI releases LFM3: six instances spanning text, audio, vision, omni — FinancialAd1961 · 2026-10-28
- Qwen3.8 Flash on a single DGX Spark gets up to 62.7% faster code generation in new update — EAccelerate_42 · 2026-10-28
- Arena co-founder: US still leads frontier AI, but China is winning open models and driving down cost — JaynitMakwana · 2026-10-28