--- title: "Drex for browser use: every click is one decision" description: "Drex Max reads the screenshot and picks the next click in one pass. 74.5 on the public Decision Index Vision against 63.9, 174 ms per screenshot decision, every screenshot size accepted." canonical: "https://nace.ai/drex/browser-use" updated: 2026-10-22 --- # Drex for browser use: every click is one decision > Drex Max reads the screenshot and picks the next click in one pass. 74.5 on the public Decision Index Vision against 63.9, 174 ms per screenshot decision, every screenshot size accepted. ## Drex for browser use. A computer-use agent makes dozens of decisions per task, each one from a screenshot. Drex Max makes each in one pass, from screens, charts, forms, photos and video, and returns a probability for every option you give it. - 74.5 Decision Index Vision, vs 63.9 - 174 ms per screenshot, vs 253 ms - 0.780 OSWorld-Verified, vs 0.734 Drex Max finding the cheapest nonstop New York to San Francisco round trip on Google Flights. Every click is one decision. ## One screenshot in. One probability per option out. The agent you already run keeps its planner and its tools. Drex Max takes the small decisions in between. 1. Send the screen. A screenshot, a screencast frame, a chart or a photo. Drex Max accepts every size. Jev 1.14 rejects 15.9% of images by side length. 2. Ask typed questions. Which element to click. Whether the page has finished loading. Which result to open. Each question lists its options, and one forward pass scores them all. 3. Act on the confident calls. Every answer is a calibrated probability. Threshold it: click when the model is sure, hand the step to a person when it is not. Median 174 ms per decision. ## Measured on screens. The public Decision Index Vision suite. A browser agent reads charts, forms, menus and photos as often as plain text. Drex Max is scored on all of them. | Benchmark | Drex Max | GPT-6 Luna Decisions | | --- | ---: | ---: | | Decision Index Vision, public | 74.5 | 63.9 | | CV-Bench | 86.3 | 73.8 | | BLINK | 74.1 | 63.7 | | RealWorldQA | 84.0 | 75.0 | | Winoground | 93.0 | 83.8 | | CharXiv | 86.1 | 78.8 | | InfographicVQA | 98.8 | 93.8 | | KIE (CORD + FUNSD) | 95.5 | 97.5 | | Mind2Web | 88.6 | 81.2 | | R-Bench-M | 48.9 | 32.5 | | MMMU-Pro (vision) | 53.6 | 41.2 | | Hateful Memes (moderation) | 70.6 | 71.2 | ### And the agent benchmarks - OSWorld-Verified (automation): 0.780 vs 0.734, desktop tasks, end to end - OSWorld-V2 (automation): 0.737 vs 0.687, desktop tasks, second edition - Mind2Web: 88.6 vs 81.2, web tasks from natural-language instructions - Screenshots accepted: 100%. Jev 1.14 rejects images by side length; Drex Max answers them all. Vision scores are Drex Max against GPT-6 Luna Decisions on the public Decision Index Vision suite, edition 0.3.2. OSWorld scores are Drex Max against Jev 1.14; Mind2Web is against GPT-6 Luna Decisions. Latency is the median API time per decision, one request at a time, on the same 243 screenshot questions from the same client. ## Fast and cheap enough for every step. A generative model pays for every token it writes. Drex Max writes nothing, so a decision costs one pass over the input no matter how many options it scores. That makes it cheap enough to call on every step of every task: before each click, after each page load, on every frame that matters. ## Two ways in. One key. Model id `drex-max`, pinned. Send the screenshot in the `images` array, up to 8 per request, with the same typed questions every Drex model takes. Priced at $0.06 per million input tokens; output is not billed. - [Open Console](https://console.nace.ai?utm_campaign=drex-max-launch&utm_content=browser_use&utm_medium=website&utm_source=nace.ai) - [OpenRouter](https://openrouter.ai/nace-ai) - [Drex Max guide](https://console.nace.ai/docs/drex/guides/drex-max) [Use cases](https://nace.ai/use-cases) ## For agents - [Drex API guide](https://nace.ai/drex.md): the decision model: request shape, question types, benchmarks and deployment, as markdown - [Decision endpoint](https://console.nace.ai/v1/systemone): POST with model "drex-latest", a state string and typed questions (choice, noul, score); https://console.nace.ai/v1/models lists checkpoints - [Document endpoints](https://console.nace.ai/v1/documents/{parse|split|classify|extract|ground}): POST a source (a public URL, an uploaded workspace file or a finished parse job) and poll GET https://console.nace.ai/v1/documents/jobs/{id}; the same key and credit as decisions - [Console](https://console.nace.ai): API keys, the playground, usage and billing; send `Authorization: Bearer $NACE_API_KEY` on every request - [Console docs](https://console.nace.ai/docs): quickstart, question types, the document routes, errors, limits and pricing - [Nace Document Intelligence](https://nace.ai/perception.md): the Perception series: parse, split, classify, extract and ground any file type with a location on every value - [Drex for Search](https://nace.ai/drex/search.md): Drex for Search: decision-powered retrieval for agents, Drex finds the page, cited search through the CLI, SDKs, MCP and REST, and the public benchmark ranks - [Use cases](https://nace.ai/use-cases.md): every Drex use case with its proof figures: browser use and search for developers, enterprise finance teams and professional firms - [Developers](https://nace.ai/developers.md): The execution layer for agents: Perception, Decision and Search behind one key, the Drex Max frontier, install steps, a scripted agent session and the four deployment boundaries - [Site index](https://nace.ai/llms.txt): every page on nace.ai; append `.md` to any page URL or send `Accept: text/markdown` for the markdown version