BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases15

LiveBench Browser

LiveBench 2.0 browser-use suite.

As of 21 Oct 2026, Claude Fable 5.1 leads LiveBench Browser on BenchLeader with 100.0%, ahead of Claude Opus 4.8 at 96.0%, across 48 model configurations with a published result.

Published by
LiveBench
Category
Agents & tools
Index weight
Reference only
Models
48
Data as of
21 Oct 2026

LiveBench (livebench.ai), White et al. 2024.

What the test looks like

LiveBench's browser-use suite: the model drives a browser to complete tasks on the open web.

How it is scored

Mean task score for the suite, 0–100, from a single run.

What to keep in mind

Sensitive to site changes and tool wiring; scores can swing between releases for the same model.

Top 15
  1. 1Claude Fable 5.1100.0%
  2. 2Claude Opus 4.896.0%
  3. 3Claude Fable 592.0%
  4. 4Gemini 4 Argon92.0%
  5. 5MiniMax-M3.1-Flash92.0%
  6. 6GPT-6 Sol92.0%
  7. 7MiMo-V2.6-Pro88.0%
  8. 8GPT-6.1 Sol88.0%
  9. 9GPT-5.6 Sol88.0%
  10. 10GPT-6 Astra84.0%
  11. 11Claude Opus 584.0%
  12. 12GLM 5.3 Flash84.0%
  13. 13GPT-5.6 Terra84.0%
  14. 14Claude Sonnet 5.584.0%
  15. 15DeepSeek V4.1 Flash84.0%
48 of 48
#
1Claude Fable 5.1100.0%66.22026-10-14
2Claude Opus 4.896.0%60.12026-10-14
3Claude Fable 592.0%67.12026-10-14
4Gemini 4 Argon92.0%63.32026-10-14
5NewMiniMax-M3.1-Flash92.0%61.92026-10-14
6GPT-6 Sol92.0%–2026-10-14
7MiMo-V2.6-Pro88.0%64.12026-10-14
8GPT-6.1 Sol88.0%63.32026-10-14
9GPT-5.6 Sol88.0%55.92026-10-14
10GPT-6 Astra84.0%67.92026-10-14
11Claude Opus 584.0%65.42026-10-14
12GLM 5.3 Flash84.0%62.72026-10-14
13GPT-5.6 Terra84.0%52.62026-10-14
14Claude Sonnet 5.584.0%–2026-10-14
15DeepSeek V4.1 Flash84.0%–2026-10-14
16Claude Opus 5.580.0%61.92026-10-14
17Muse Spark 1.380.0%58.32026-10-14
18Claude Haiku 5.580.0%–2026-10-14
19NewDeepSeek V4.1 Flash 101776.0%–2026-10-14
20Muse Wonder72.0%63.62026-10-14
21Claude Sonnet 572.0%56.02026-10-14
22GPT-6 Luna72.0%–2026-10-14
23Claude Sonnet 4.668.0%57.12026-10-14
24GPT-5.6 Luna68.0%47.92026-10-14
25Grok 4.664.0%57.82026-10-14
26Gemini 3.8 Flash64.0%55.62026-10-14
27Deepseek v4 Flash Vision64.0%–2026-10-14
28Grok 4.760.0%–2026-10-14
29Gemini 3.7 Flash56.0%53.22026-10-14
30GPT-5.548.0%60.22026-10-14
31Qwen3.8 Flash48.0%–2026-10-14
32Grok 4.544.0%58.92026-10-14
33Claude Haiku 4.540.0%45.32026-10-14
34Muse Spark 1.236.0%–2026-10-14
35NewMuse Spark Open36.0%–2026-10-14
36Qwen3.8 Maxmax32.0%63.32026-10-14
37Qwen3.8 27B32.0%54.52026-10-14
38Step 5 Preview24.0%60.62026-10-14
39Kimi K316.0%59.12026-10-14
40Mistral Large 48.0%56.72026-10-14
41MiniMax M38.0%56.52026-10-14
42Gemini 3.1 Pro4.0%60.72026-10-14
43Gemma 4 31B4.0%48.32026-10-14
44Nemotron Super VL 3.54.0%–2026-10-14
45NewBeam 1.00.0%54.62026-10-14
46GLM 5.30.0%–2026-10-14
47Nemotron Lightning 3.50.0%–2026-10-14
48Nemotron Ultra 30.0%–2026-10-14

More agents & tools benchmarks · All LLM benchmarks

Cite as: BenchLeader, “LiveBench Browser leaderboard”, https://www.benchleader.com/benchmarks/livebench_browser, data as of 21 Oct 2026.

LiveBench Browser: questions

What does LiveBench Browser measure?
LiveBench's browser-use suite: the model drives a browser to complete tasks on the open web. Scores are reported in percent of tasks solved; higher is better.
Which AI model leads LiveBench Browser?
Claude Fable 5.1 leads LiveBench Browser with 100.0% as of 21 Oct 2026, ahead of Claude Opus 4.8 at 96.0%.
How many models have LiveBench Browser results?
48 model configurations have a LiveBench Browser result on BenchLeader, all taken from LiveBench.
Who runs LiveBench Browser and how often is it updated?
LiveBench Browser is published by LiveBench. BenchLeader re-reads the published results once a day and records the date each result was published.
Does LiveBench Browser count toward the BenchLeader Index?
No. LiveBench Browser is shown for reference but left out of the composite index.