gpt-6IMAGE
RSI-Exam leaderboard: Opus 5.5 overtakes GPT-6 Astra for first place
On 88 executable research tasks, Claude Opus 5.5 scored 0.5357 to take first, GPT-6 Astra placed second at 0.5126, and Grok 4.7 rose to 0.3842.
Huaxiu Yao
@HuaxiuYaoMLArchived 2026.10.03Originally posted 2026.10.02
Huaxiu Yao @HuaxiuYaoMLOriginally posted 2026.10.02
🚨 New #1 on RSI-Exam 🏆 🥇 Claude Opus 5.5 from @AnthropicAI: 0.5357 across all 88 executable research tasks 🥈 GPT-6-astra: 0.5126 📊 Opus 5.5 leads on full, public & private splits 🆕 Grok 4.7 @SpaceXAI is up: 0.3842 (vs 0.3671 for Grok 4.6) 📈 ⏳ Gemini 4 is on the way 👀 🔗 https://rsi-exam.ai/
x.com/huaxiuyaoml/status/2105880459133800639Archived 2026.10.03 10:05