{"slug": "google-deepmind-and-artificial-analysis-launch-mka-benchmark-for-visual-ai", "title": "Google DeepMind and Artificial Analysis launch MKA benchmark for visual AI agents", "summary": "Google DeepMind and Artificial Analysis released the Multimodal Knowledge Agent benchmark (MKA) on November 10, 2026, grading autonomous agents on 200 private tasks across 10 industry verticals built with input from specialists representing 83 occupations. The benchmark requires agents to interpret architectural blueprints, engineering schematics and clinical scans and produce deliverables such as spreadsheets, slide decks and marked-up images on projects that take human professionals 3 to 8 hours, scored by expert rubrics and an agentic autorater that reached 96.2% agreement with human experts. On the first leaderboard, Gemini 4 Carbon (high) ranked second in quality while completing tasks at roughly 6x lower cost than the top-ranked model, a result DeepMind's co-authorship of the benchmark warrants weighing.", "body_md": "# Google DeepMind and Artificial Analysis launch MKA benchmark for visual AI agents\n\nPhoto: Steve A Johnson / Pexels\n\nThe new multimodal knowledge agent benchmark tests whether AI can handle blueprints, schematics and clinical scans like a working professional\n\n[Google](https://cryptobriefing.com/markets/alphabet/) DeepMind and Artificial Analysis have released a new yardstick for AI agents, and it involves a lot more squinting at diagrams than most benchmarks do.\n\nThe Multimodal Knowledge Agent benchmark, or MKA, grades autonomous agents on complex, visually heavy professional work. It skips trivia questions and puzzle-solving. It is closer to a performance review.\n\n## What MKA actually tests\n\nThe two organizations unveiled MKA on November 10, 2026. The release coincided with a DeepMind blog post and an accompanying technical paper.\n\nThe benchmark contains 200 private tasks spread across 10 industry verticals. Each task was shaped with input from specialists representing 83 different occupations.\n\nAgents must interpret architectural blueprints, engineering schematics, clinical scans and other dense visual material. Agents then have to produce real deliverables, including spreadsheets, slide decks and marked-up images.\n\nThe tasks are modeled on multi-step projects that typically take human professionals 3 to 8 hours to complete.\n\nScoring follows strict expert rubrics. Those rubrics weigh factual accuracy alongside visual design, so a correct answer buried in an ugly, unreadable slide deck still loses points.\n\n## Grading the graders\n\nMKA leans on an agentic autorater, an AI system that scores the outputs of other AI systems.\n\nAccording to the research findings, the autorater reached a 96.2% agreement rate with human experts on the evaluation criteria.\n\n### AI, tech, and the markets they move—in one daily briefing.\n\nDaily. Free. Join 34,000+ readers across crypto, finance, and policy.\n\nThe private nature of the tasks matters too. Because the 200 tasks are not public, developers cannot easily train their models on the answers ahead of time.\n\n## Early results: quality versus cost\n\nThe first leaderboard already contains a notable data point for Google. Gemini 4 Carbon (high) ranked second in quality on MKA.\n\nThe more eye-catching figure is efficiency. Gemini 4 Carbon (high) completed tasks at roughly 6x lower cost than the model that took the top spot.\n\nIt is worth noting the obvious. DeepMind co-built the benchmark, and its own model scored well on cost-adjusted performance. That does not invalidate the results, but readers should weigh them with that relationship in mind.\n\n## Why visual reasoning is the missing piece\n\nMany existing AI benchmarks rely on text-only tasks. MKA is designed to fill that gap by pulling visual reasoning into the evaluation itself.\n\nThat gap matters because professional work is rarely pure text. Architects work from drawings, engineers from schematics, and clinicians from scans.\n\nBy requiring both visual comprehension and polished output, MKA tests the full loop of knowledge work: take in complex material, reason about it, and hand back a usable product.\n\n## What this means for AI developers and buyers\n\nFor model developers, MKA raises the stakes on multimodal capability. It also puts design quality on the scorecard. Developers can no longer treat formatting as an afterthought if expert rubrics penalize cluttered or confusing deliverables.\n\nTwo hundred tasks across 10 sectors is substantial for a hand-curated benchmark, but it still covers a slice of professional work, and results in one vertical may not transfer neatly to another.\n\nThe autorater’s 96.2% agreement with human experts is strong, yet the remaining disagreement could cluster in the hardest or most subjective tasks.\n\nArtificial Analysis brings a third-party evaluation brand to the project, while DeepMind brings the research muscle and a direct competitive interest. How other labs respond, and whether they submit models and accept the methodology, will determine whether MKA becomes an industry standard or remains one benchmark among many.\n\n**Disclosure:** This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our [Editorial Policy](https://cryptobriefing.com/editorial-policy/).", "url": "https://wpnews.pro/news/google-deepmind-and-artificial-analysis-launch-mka-benchmark-for-visual-ai", "canonical_source": "https://cryptobriefing.com/deepmind-artificial-analysis-mka-benchmark/", "published_at": "2026-11-10 18:21:17+00:00", "updated_at": "2026-11-10 18:51:17.401409+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-agents", "ai-research", "large-language-models", "ai-safety"], "entities": ["Google DeepMind", "Artificial Analysis", "Multimodal Knowledge Agent benchmark", "Gemini 4 Carbon"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/google-deepmind-and-artificial-analysis-launch-mka-benchmark-for-visual-ai", "markdown": "https://wpnews.pro/news/google-deepmind-and-artificial-analysis-launch-mka-benchmark-for-visual-ai.md", "text": "https://wpnews.pro/news/google-deepmind-and-artificial-analysis-launch-mka-benchmark-for-visual-ai.txt", "jsonld": "https://wpnews.pro/news/google-deepmind-and-artificial-analysis-launch-mka-benchmark-for-visual-ai.jsonld"}}