Understand this story
Artificial Analysis has launched the CritPt v2 benchmark. According to the information it released, 70 research-level physics problems were fully reviewed by the authors' team, and the answering method was changed to an agentic approach completed in an offline sandbox with the help of code execution tools. The leaderboard shows that GPT-6 Astra (max) and GPT-6.1 Sol (max) rank first with 83% and 82% respectively, Claude Opus 5.5 (max with fallback) is at 78%, and GPT-5.6 Sol (max) is at 75%. GPT-5 (high), released in August 2025, scored only 29%, 54 percentage points lower than GPT-6 Astra (max). The benchmark is currently online, with no subsequent updates yet.
AI-generated from public reports
Latest development · Report
CritPt v2 launches on Artificial Analysis: GPT-6 Astra tops the leaderboard at 83%Developments
1 confirmed developments
- Source:Artificial Analysis
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Oct 23
Artificial Analysis
CritPt v2 launches on Artificial Analysis: GPT-6 Astra tops the leaderboard at 83%
CritPt v2 is now live on Artificial Analysis. All 70 research-level physics problems have been fully reviewed by the authors' team, and the benchmark has been changed to agentic answering in an offline sandbox with code execution tools. GPT-6 Astra (max) and GPT-6.1 Sol (max) lead at 83% and 82% respectively, followed closely by Claude Opus 5.5 (max with fallback) at 78% and GPT-5.6 Sol (max) at 75%. GPT-5 (high), released in August 2025, scores only 29%, 54 percentage points below GPT-6 Astra (max).