Claude Opus 4.7
Anthropic · 40.9%
0255075100
Home · source disagreements
Where the public record
splits, on purpose.
The product should make the public record easier to read, especially when independent sources disagree loudly.
Build / data stamp
Read this before trusting a headline.
Data snapshot May 1, 2026Registry verification passed9 providers · 826 tracked modelsPage refreshed May 7, 2026
If this stamp lags behind the repo, you are likely looking at an older build or cached deploy.
Disagreement view · spread deskRankings get worse more honest
Rankings get worse more honest
when spread stays visible.
This view surfaces where public sources refuse to tell the same story. A wide spread is not noise to be sanded off. It is the main fact.
ArenaAR
LiveBenchLB
Artificial AnalysisAA
BridgeBenchBB
Terminal-BenchTERMINAL-BENCH
LLMBaseLLMBASE
Scale LabsSL
OpenCompassOC
MTEBMTEB
Where the public record splits
cross-benchmark spread = max percentile − min percentileGemini 2.5 Pro
Google · 56.8%
0255075100
Gemini 3.1 Pro
Google · 29.5%
0255075100
GPT-5.4
OpenAI · 50%
0255075100
GPT-5.5
OpenAI · 43.2%
0255075100
GPT-5.4 mini
OpenAI · 27.3%
0255075100
DeepSeek Reasoner
DeepSeek · 27.3%
0255075100
Source honesty scorecard
Not a moral rating. A quick check on how inspectable each source is when you need to dispute the headline number.
9 of 9 sources in the current registry
Benchmark and eval counts reflect what this app currently tracks for each source, not the source's full external catalog.
| Arena verified | 11 | 773 | no | May 1, 2026 | 1 |
| Artificial Analysis verified | 7 | 690 | yes | May 1, 2026 | 1 |
| LiveBench verified | 6 | 192 | yes | May 1, 2026 | 1 |
| Scale Labs verified | 8 | 98 | no | May 1, 2026 | 0 |
| BridgeBench verified | 5 | 52 | no | May 1, 2026 | 1 |
| Terminal-Bench verified | 1 | 24 | no | May 1, 2026 | 1 |
| OpenCompass verified | 1 | 15 | no | May 1, 2026 | 0 |
| MTEB verified | 1 | 1 | no | May 1, 2026 | 0 |
| LLMBase relay | 0 | 0 | no | May 1, 2026 | 0 |