// AI news · 2026-11-11 · Google Gemini
Google Launches Multimodal Knowledge Agent Benchmark
Google and Artificial Analysis launch the Multimodal Knowledge Agent (MKA) benchmark to evaluate agents' reasoning on visually complex files.
Published 2026-11-11 00:00 (Beijing time)Source Google GeminiAuto-summarised, fact-checked against source
01Summary
It includes 200 real-world tasks designed by experts across 10 industries and 83 occupations, each with multimodal inputs and expert rubrics. Models vary widely in performance, using between 30 and 130 turns per task. Gemini 4 Carbon achieves top quality at 6x lower cost than the best-performing model and is available via Google AI Studio.
Summarised from the original article by an open-source model, then checked automatically: every number and English name must appear in the source. The original article is authoritative.
02Key points
- MKA benchmark includes 200 real-world tasks from 83 occupations across 10 industries.
- Each task requires 3–8 hours of expert work and features multimodal inputs and rubrics.
- Models vary significantly in turn efficiency, from 30 to 130 turns per task.
- Gemini 4 Carbon achieves top quality at 6x lower cost than the leading model.
03Source
Introducing the Multimodal Knowledge Agent benchmark
Google Gemini · published 2026-11-11 00:00 Beijing time
05Related news
- Same source · Gemini connects to Google Wallet for finance insights2026-11-14 · Google Gemini
- Same source · Gemini lets you generate and edit docs in chat2026-10-30 · Google Gemini
- Same topic · Experimenting with CNP in GKE for Network Policy Control2026-11-11 · Google Cloud
- Same topic · Gemini 4 Carbon Now Available2026-11-11 · Google Keyword
- Same day · Gemini 4 Carbon Launches in Gemini App2026-11-11 · 9to5Google
From AIMap AI news: collected from 1,100+ official and media sources. Summaries only restate the original article; follow the link for the full story. All stories of 2026-11-11 · Archive by date · All AI news · Today