Best LLM for web browsing agents
- Claude Fable 5.1 leads
- 100.0%
- models ranked
- 43
- lead over #2
- +4.0 pts
LiveBench browser · read October 22
As of October 22, 2026, Anthropic Claude Fable 5.1 leads 43 models on LiveBench browser with 100.0%, ahead of Anthropic Claude Opus 4.8 (96.0%) and Anthropic Claude Fable 5 (92.0%).
The full best LLM for agents ranking combines this benchmark with others.
| Rank | Model | LiveBench browser | Coding evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Claude Fable 5.1 Anthropic | 100.0% | 84.2 | $50.00 | |
| 2 | Claude Opus 4.8 Anthropic | 96.0% | 65.0 | $25.00 | |
| 3 | Claude Fable 5 Anthropic | 92.0% | 79.3 | $50.00 | |
| 4 | Gemini 4 Argon Google | 92.0% | 81.0 | $10.00 | |
| 5 | GPT-6 Sol OpenAI | 92.0% | 76.0 | $10.00 | |
| 6 | GPT-5.6 Sol OpenAI | 88.0% | 73.5 | $20.00 | |
| 7 | GPT-6.1 Sol OpenAI | 88.0% | 82.4 | $10.00 | |
| 8 | MiMo-V2.6-Pro Xiaomi | 88.0% | 65.4 | $0.87 | |
| 9 | Claude Opus 5 Anthropic | 84.0% | 83.0 | $25.00 | |
| 10 | Claude Sonnet 5.5 Anthropic | 84.0% | 87.3 | $10.00 | |
| 11 | DeepSeek V4.1 Flash DeepSeek | 84.0% | 62.5 | $0.84 | |
| 12 | GPT-5.6 Terra OpenAI | 84.0% | 63.9 | $12.00 | |
| 13 | GPT-6 Astra OpenAI | 84.0% | 85.1 | $50.00 | |
| 14 | GLM 5.3 Flash Z.ai | 84.0% | 57.7 | $0.50 | |
| 15 | Claude Haiku 5.5 Anthropic | 80.0% | 70.7 | $0.50 | |
| 16 | Claude Opus 5.5 Anthropic | 80.0% | 88.8 | $20.00 | |
| 17 | Muse Spark 1.3 Meta | 80.0% | 57.4 | $4.25 | |
| 18 | DeepSeek V4.1 Flash 1017 DeepSeek | 76.0% | 70.3 | $0.60 | |
| 19 | Claude Sonnet 5 Anthropic | 72.0% | 57.7 | $10.00 | |
| 20 | GPT-6 Luna OpenAI | 72.0% | 57.3 | $0.50 |
How this list is built
Generated from the dataset on every refresh. Nobody picks the order by hand.
Models by their LiveBench browser result, as the board publishes it, best first. Multi-step tasks in working web applications, done through a real browser; a model that cannot take screenshots scores 0. Each score is the model's mean task score for the category, 0 to 100, from a single run of the current release, shown as a share.
- OpenRouter, fetched
- LiveBench, fetched , board updated October 14, 2026