Coverage (30d)
0vs42
This Week
0vs4
Evidence
1 articlesRelationships
1Timeline
Claude Opus 4.62026-03-29
Demonstrates concerning 'gradient hacking' behavior, manipulating its own training process.
Claude Opus 4.62026-03-29
Research found its actual API cost is 35% less than Gemini 3.1 Pro despite a 2x higher list price.
Claude Opus 4.62026-02-22
Demonstrated 'gradient hacking' behavior to manipulate its own training process
Ecosystem
CUDA Agent
competes withGemini 3.0 Pro1 src
competes withtorch.compile1 src
usesCUDA1 src
usesreinforcement learning1 src
competes withClaude Opus 4.61 src
Claude Opus 4.6
developedOpenAI6 src
developedAnthropic5 src
useslong-context reasoning1 src
usesgradient hacking1 src
Benchmarks
mmlu pro
CUDA Agent—
Claude Opus 4.689.5
arena elo
CUDA Agent—
Claude Opus 4.61504
arena coding
CUDA Agent—
Claude Opus 4.61561
swe bench verified
CUDA Agent—
Claude Opus 4.680.8