All rankings / Search and documents

Search

Human preference ratings for the overall category of this Arena task. Scores from different boards are not interchangeable.

42 entries · Arena score · Higher is better

Source publication: 2026-11-09. Retrieved 2026-11-12.

Source: Arena (lmarena-ai) · CC-BY-4.0.

Ranks use confidence bounds. Several models can share a rank; the rows within a shared rank are displayed by score.

Search · Arena score
Significance rankModel / evaluated variantArena score95% confidence intervalVotes
1Qwen3.8-Maxalibaba1,259.781,246.19 – 1,273.381,998
1GPT-5.6 Sol (xHigh)openai1,257.841,251.39 – 1,264.2962,364
1Claude Opus 4.6 (Search)anthropic1,251.671,247.03 – 1,256.32166,732
2GPT-5.5 (Search)openai1,245.861,240.69 – 1,251.03121,489
4Claude Opus 5 (High)anthropic1,239.031,232.23 – 1,245.8234,000
4Claude Fable 5.1 (High)anthropic1,237.311,230.41 – 1,244.233,769
5GPT-6 Astra (Max)openai1,230.811,222.94 – 1,238.6817,456
5GPT-6.1 Sol (Max)openai1,230.361,222.57 – 1,238.1626,556
5Claude Opus 4.7anthropic1,230.11,225.11 – 1,235.09121,890
5Claude Fable 5 (High)anthropic1,228.151,221.51 – 1,234.869,528
5ERNIE 5.1baidu1,226.81,217.21 – 1,236.393,925
7Claude Sonnet 4.6 (Search)anthropic1,220.621,215.99 – 1,225.25167,499
10Grok 4.5xai1,215.821,210.07 – 1,221.57105,282
11Claude Opus 5.5 (Medium)anthropic1,213.331,206.55 – 1,220.1131,730
13Gemini 3.1 Pro (Grounding)google1,210.41,205.78 – 1,215.02171,494
13GPT-5.2 (Search)openai1,207.081,201.45 – 1,212.754,658
13Gemini 3 Pro (Grounding)google1,206.571,201.02 – 1,212.1238,324
13Grok 4.20 Multi-Agent Beta (0309)xai1,205.161,199.61 – 1,210.7121,069
14Claude Opus 4.8anthropic1,202.421,196.65 – 1,208.18101,197
16GPT-5.1 (Search)openai1,199.421,194.15 – 1,204.6861,989
16GPT-5.4 (Search)openai1,198.031,192.47 – 1,203.59121,580
18Gemini 3 Flash (Grounding)google1,195.571,191.16 – 1,199.99189,590
19Grok 4.20 Beta 1xai1,190.881,184.84 – 1,196.9355,902
20Claude Sonnet 5 (Search)anthropic1,187.741,181.31 – 1,194.1871,297
23Claude Opus 4.5 (Search)anthropic1,179.171,173.51 – 1,184.8463,904
24GPT-6 Sol (Max)openai1,174.471,167.02 – 1,181.9213,499
25GPT-5.2 (Search, Non-Reasoning)openai1,171.871,166.28 – 1,177.4678,443
25Grok 4.1 Fast (Search)xai1,171.631,166.62 – 1,176.6484,455
25Grok 4 Fast (Search)xai1,170.631,165.91 – 1,175.3542,640
26Grok 4.3xai1,166.431,161.59 – 1,171.27122,703
26GPT-6 Luna (Max)openai1,160.171,153.23 – 1,167.146,776
30Claude Sonnet 4.5 (Search)anthropic1,159.191,154.01 – 1,164.37139,468
33Claude Opus 4.1 (Search)anthropic1,148.191,143.25 – 1,153.1478,777
33o3 (Search)openai1,144.531,138.85 – 1,150.2120,655
33Gemini 2.5 Pro (Grounding)google1,143.391,138.63 – 1,148.1685,505
33Grok 4 (Search)xai1,141.881,135.72 – 1,148.0519,069
33Sonar Reasoning Pro (High)perplexity1,138.811,132.91 – 1,144.7229,197
34GPT-5 (Search)openai1,133.091,127.03 – 1,139.1520,742
37Sonar Pro (High)perplexity1,129.71,123.8 – 1,135.5928,684
38Claude Opus 4 (Search)anthropic1,127.041,121.55 – 1,132.5431,275
41Diffbot-Small-XLdiffbot1,022.041,013.47 – 1,030.616,339
41GPT-4o (Search)openai1,006.36995.1 – 1,017.633,392

What this ranking can tell you

  • The population is the models in this source board, including variants and models without an available API price.
  • Ranks use the published confidence bounds: one plus the number of models whose lower bound exceeds this model’s upper bound. Shared ranks are not proof of equal performance.

Check the source

The original evidence is retained byte for byte. The downloadable board includes the source URLs, revision and SHA-256 hashes.

Download this board and provenance →

Source files and licences

Explore other model types →Read the benchmark guide →

Evidence & Ask