New: Claude Fable 5.2 eval report is here.Read the report ↗
Compare models

Best models for finding and fixing bugs

SWE-sweep

Tests autonomous discovery and repair of bugs in repositories without issue descriptions. This reported comparison retains differing agent harnesses.

Updated: Data as of:
Limited reportingNot saturated
Official benchmark
SWE-sweep results. Each row identifies its benchmark version and evaluation track. Scores from different versions are not directly comparable. Best effort from the latest stored report date per model and version.
Model / creatorVersionReported score
Gemini 4 HydrogenGoogle · ProprietaryGoogle Hydrogen launch / Antigravity CLI 1.4.813.5%
GPT-6.1 TerraOpenAI · ProprietaryGoogle Hydrogen launch / Antigravity CLI 1.4.85.8%
Claude Sonnet 5.5Anthropic · ProprietaryGoogle Hydrogen launch / Antigravity CLI 1.4.83.1%
View all 3 published reports
Google Hydrogen launch / Antigravity CLI 1.4.8

effort: highest available or best reported · Hydrogen uses Antigravity CLI 1.4.8 with safety filters; peer figures are quoted from the public leaderboard. · Tools enabled

Google DeepMind ↗
Google Hydrogen launch / Antigravity CLI 1.4.8

effort: highest available or best reported · Hydrogen uses Antigravity CLI 1.4.8 with safety filters; peer figures are quoted from the public leaderboard. · Tools enabled

Google DeepMind ↗
Google Hydrogen launch / Antigravity CLI 1.4.8

effort: highest available or best reported · Hydrogen uses Antigravity CLI 1.4.8 with safety filters; peer figures are quoted from the public leaderboard. · Tools enabled

Google DeepMind ↗

Sources and original reports ↗