
Design and engineering consultancy for devtools, AI, and cybersecurity startups. Lago, Tines, Teleport, and Whop grow with Evil Martians. We host @sfrubyconf.
- Tweets4.8K
- Following190
- Followers9.9K
- Likes3.1K

ALT “Agents on Rails” benchmark results from the Rails Foundation, testing completed November 2026. Thirteen AI model configurations are ranked by accuracy on 46 Rails coding tasks, with average time, tokens and cost per run. GPT-6.2 Astra (xhigh) leads at 65.8%, followed by Claude Opus 5.5 (xhigh) at 65.3% and Gemini 4 Argon (high) at 60.6%. Four newly tested configurations are highlighted with a red glow: GPT-6.2 Astra (xhigh and medium), GPT-6.2 Sol (xhigh) and DeepSeek V4.1 Flash 1017 (max).

ALT “Agents on Rails” benchmark results from the Rails Foundation, testing completed November 2026. Thirteen AI model configurations are ranked by accuracy on 46 Rails coding tasks, with average time, tokens and cost per run. GPT-6.2 Astra (xhigh) leads at 65.8%, followed by Claude Opus 5.5 (xhigh) at 65.3% and Gemini 4 Argon (high) at 60.6%. Four newly tested configurations are highlighted with a red glow: GPT-6.2 Astra (xhigh and medium), GPT-6.2 Sol (xhigh) and DeepSeek V4.1 Flash 1017 (max).















