Updated daily at 05:00 UTC
Benchmark
Search Arena (arena.ai)
Measures: Users ask questions to two anonymous search-augmented (web-grounded) models and vote for the better cited answer; votes are turned into an Elo-scale score with a confidence interval (no adjustment by default; Style Control and Factuality optional).
- Maintainer
- Arena Intelligence (arena.ai; formerly LMArena / LMSYS Chatbot Arena)
- Introduced
- 2025
- Status
- Active
- Saturation
- Headroom
- Typical provenance
- Independent
- In scoring set
- Yes
Leaderboard
Every result we hold, whoever ran it. Only independent and third-party rows count towards a score.
| Model | Arena score (Elo scale) | Score CI (+/-) | Votes | Provenance | Source | Evaluated |
|---|---|---|---|---|---|---|
| GPT-5.6 Sol OpenAI · Xhigh effort | 1256 | — | 29663 | Independent | Arena | 24 Aug 2026 |
| Claude Fable 5 | 1229 | — | 41795 | Independent | Arena | 24 Aug 2026 |
| Claude Opus 4.7 | 1212 | — | 91394 | Independent | Arena | 24 Aug 2026 |
| Grok 4.20 Multi-Agent | 1205 | — | 109553 | Independent | Arena | 24 Aug 2026 |
| Grok 4.5 | 1202 | — | 31505 | Independent | Arena | 24 Aug 2026 |
| Claude Opus 4.8 | 1195 | — | 70998 | Independent | Arena | 24 Aug 2026 |
| ERNIE 5.1 | 1193 | — | 3788 | Independent | Arena | 24 Aug 2026 |
| Grok 4.3 | 1193 | — | 91483 | Independent | Arena | 24 Aug 2026 |
- Arena (arena.ai), lmarena-ai/leaderboard-dataset, CC BY 4.0.