One question dominates every conversation we have with developers: _"Which model should I plug into Browser Use?"_ A single test run means nothing on the chaotic, ever-changing real web. We run our benchmark multiple times per model and aggregate with statistical bootstrapping for real error bars. Our [evaluation system](/content/posts/sota-technical-report/index.html) has run over 600,000 tasks, validated by an LLM judge achieving 87% agreement with human labels. ## The results [Link to The results](/content/posts/what-model-to-use#the-results/index.html) | Model | Type | Score | | --- | --- | --- | | **Browser Use Cloud (bu-ultra)** | Cloud | **78.0%** | | OSS + BU LLM (ChatBrowserUse-2) | OSS + Cloud LLM | 63.3% | | claude-opus-4-6 | Open Source | 62.0% | | gemini-3-1-pro | Open Source | 59.3% | | claude-sonnet-4-6 | Open Source | 59.0% | | gpt-5 | Open Source | 52.4% | | gpt-5-mini | Open Source | 37.0% | | gemini-2.5-flash | Open Source | 35.2% | "Open Source" means running the open-source Browser Use library with that LLM. "OSS + Cloud LLM" means the open-source library with our ChatBrowserUse-2 model. "Cloud" is the fully managed Browser Use Cloud agent. You can view all models and their prices [here](https://docs.browser-use.com/supported-models#available-models). 100 hand-selected tasks from five sources (Custom, WebBench, Mind2Web 2, GAIA, BrowseComp). Each task is hard but verified completable. Full methodology in our [benchmark post](/content/posts/ai-browser-agent-benchmark/index.html). ## The model breakdown [Link to The model breakdown](/content/posts/what-model-to-use#the-model-breakdown/index.html) ### Browser Use Cloud (`bu-ultra`) -- 78.0% [Link to Browser Use Cloud (bu-ultra) -- 78.0%](/content/posts/what-model-to-use#browser-use-cloud-bu-ultra----780/index.html) The clear winner. Not just the highest accuracy but also the fastest at ~14 tasks per hour. Each step is slower than a raw frontier LLM call, but bu-ultra completes tasks in far fewer steps, so total wall-clock time is lower. This isn't just a better model. [Introduced in Browser Use 1.0](/content/posts/speed-matters/index.html), it's a purpose-built agent with [stealth browser infrastructure](/content/posts/stealth-benchmark/index.html), CAPTCHA solving, persistent filesystem, and optimized tool orchestration. The 16-point gap over the best frontier model comes from full-stack optimization. **Verdict**: Best performance, highest throughput, zero setup. Use this. ### ChatBrowserUse-2 (OSS + Cloud LLM) -- 63.3% [Link to ChatBrowserUse-2 (OSS + Cloud LLM) -- 63.3%](/content/posts/what-model-to-use#chatbrowseruse-2-oss--cloud-llm----633/index.html) Our model specifically optimized for browser automation, running on the open-source library. It outperforms every standalone frontier model while being faster and cheaper per task. **Verdict**: Best option if you need custom tools or self-hosting but still want top-tier accuracy. ### Claude -- 62.0% (opus), 59.0% (sonnet) [Link to Claude -- 62.0% (opus), 59.0% (sonnet)](/content/posts/what-model-to-use#claude----620-opus-590-sonnet/index.html) Claude models are the strongest standalone frontier option. Claude-opus-4-6 leads all non-Browser Use models at 62%. When the agent needs to execute custom JavaScript or extract complex structured data, Claude is unparalleled. Claude-sonnet-4-6 at 59% is close behind opus at roughly half the cost, making it the best value among Anthropic models. **Verdict**: Best standalone frontier model. Use Claude when your workflow relies on custom code execution. ### Gemini -- 59.3% (3-1-pro), 35.2% (2.5-flash) [Link to Gemini -- 59.3% (3-1-pro), 35.2% (2.5-flash)](/content/posts/what-model-to-use#gemini----593-3-1-pro-352-25-flash/index.html) Gemini-3-1-pro scores 59.3%, neck and neck with Claude Sonnet. Strong vision, low latency, and massive context windows that handle enormous DOMs without issue. Gemini-2.5-flash at 35.2% is the fastest cheap option but accuracy drops hard. You get what you pay for. **Verdict**: Gemini-3-1-pro is a strong alternative to Claude. Flash is only for cost-sensitive, low-stakes tasks. ### OpenAI -- 52.4% (gpt-5), 37.0% (gpt-5-mini) [Link to OpenAI -- 52.4% (gpt-5), 37.0% (gpt-5-mini)](/content/posts/what-model-to-use#openai----524-gpt-5-370-gpt-5-mini/index.html) GPT-5 scores 52.4% and is the slowest model on the benchmark at ~6 tasks per hour. GPT-5-mini at 37% doesn't make up for the speed gap. Recent OpenAI models have not kept pace with Claude and Gemini on browser automation tasks. **Verdict**: Falling behind competitors. Use Claude or Gemini instead. ## Recommendations [Link to Recommendations](/content/posts/what-model-to-use#recommendations/index.html) - **Want the best agent?** Use [Browser Use Cloud](https://cloud.browser-use.com/) (`bu-ultra`). 78% accuracy, fastest throughput, no setup. - **Need custom tools or self-hosting?** Use the open-source library with `ChatBrowserUse-2`. 63.3%, still beats every frontier model. - **Prefer a standalone frontier LLM?** Claude-opus-4-6 (62%) or gemini-3-1-pro (59.3%). - **On a budget?** Claude-sonnet-4-6 (59%) gives near-opus accuracy at lower cost. The benchmark is open source at [github.com/browser-use/benchmark](https://github.com/browser-use/benchmark). If you're an LLM provider looking to test at scale, reach out at support@browser-use.com. [Alexander Yue](/content/team/alexander-yue?from=blog/index.html)·February 19, 2026