Every race.
One track each.
Each capability is a lap. Distance between racers is the real score gap, so a photo finish looks like one and a runaway lead looks like one too.
19
Photo finishes
11
Dominant leads
42
Lead changes (30d)
53 wk
Longest streak
Overall · live
4 weeks at #1
Claude Opus 5.5
leads by 0.1 pts over GPT-6.1 Sol
Overall
The headline race: a weighted blend of every capability.
Coding
From UI polish to distributed systems and autonomous repo work.
Hard Reasoning
Novel problems, logic, planning and expert-level questions.
Math
Competition problems through research-grade mathematics.
Science
Graduate-level science, biology, physics and research help.
Writing & Language
Creative and technical writing, editing, languages and conversation.
Multimodal
Understanding images, documents, video and audio.
Tool Use & Agents
Calling tools, driving computers, browsing and working for hours.
Long Context
Finding and reasoning over information in million-token inputs.
Speed & Cost
Throughput, responsiveness and intelligence per dollar.
Trust & Factuality
Avoiding hallucinations and knowing what it does not know.
Image Generation
Photoreal, illustration, anime, typography and editing.
Video Generation
Cinematic realism, animation, physics, sound and consistency.