Telegram

Best model for search

Answers that cite the live web, ranked on whether the citation holds up.

Right now

GPT 5.6 Sol Xhigh from OpenAI currently ranks first for answers grounded in live sources, ahead of Qwen3.8 Max from Alibaba. The ranking comes from LMArena and was last published 2026-10-13.

Leader
GPT 5.6 Sol Xhigh
Made by
OpenAI
Score
1261
Source
LMArena

Full ranking Human head-to-head preference votes, Bradley-Terry rating.

Rank# Model Made by Arena score Votes
1 GPT 5.6 Sol Xhigh OpenAI 1,261 60,437
2 Qwen3.8 Max Alibaba 1,261 1,870
3 Claude Opus 4.6 Search Anthropic 1,251 173,191
4 GPT 5.5 Search OpenAI 1,247 125,957
5 Claude Fable 5.1 High Anthropic 1,246 12,782
6 Claude Opus 5 High Anthropic 1,242 13,285
7 GPT 6 Astra Max OpenAI 1,234 18,483
8 GPT 6.1 Sol Max OpenAI 1,233 5,555
9 Claude Opus 4.7 Anthropic 1,230 127,506
10 Claude Fable 5 High Anthropic 1,229 73,596
11 Ernie 5.1 Baidu 1,227 4,024
12 Claude Sonnet 4.6 Search Anthropic 1,221 173,986
13 Grok 4.5 xAI 1,216 76,458
14 Claude Opus 5.5 Medium Anthropic 1,213 10,842
15 Gemini 3.1 Pro Grounding Google 1,209 151,436
16 GPT 5.2 Search OpenAI 1,207 55,634
17 Gemini 3 Pro Grounding Google 1,206 38,857
18 Grok 4.20 Multi Agent Beta xAI 1,205 126,238
19 Claude Opus 4.8 Anthropic 1,204 106,304
20 GPT 5.1 Search OpenAI 1,200 63,042
21 GPT 5.4 Search OpenAI 1,199 126,730
22 Gemini 3 Flash Grounding Google 1,196 196,399
23 Claude Sonnet 5 Search Anthropic 1,191 72,978
24 Grok 4.20 Beta1 xAI 1,190 57,049
25 Claude Opus 4.5 Search Anthropic 1,180 65,047

Published by LMArena on 2026-10-13 · copied here · no adjustment applied

Other leaderboards