Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Agentic benchmark
Multimodal Knowledge Agent Benchmark. Every model the catalog carries a published MKA value for, ranked by that value.
Google DeepMind's benchmark of professional knowledge work where the key information sits in visual files such as architectural drawings, engineering schematics and medical scans. The agent has to deliver a finished report, spreadsheet, slide deck or marked-up drawing. Artificial Analysis runs it on a private 200-task set. 22 models in the ModelChoice catalog have a published value.
Each point is a model with a published MKA value and complete token prices, on the 75% input / 25% output blend. Frontier models are spaced evenly along the cost axis.
22 models with a published MKA value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.
| Rank | Model | Provider | Climb WSR (weighted rubric satisfaction rate, %) |
|---|---|---|---|
| 1 | Anthropic | 56.6 | |
| 2 | 56.5 | ||
| 3 | 54.3 |
What the benchmark measures and how its value is published, from the source’s own record.
MKA (Multimodal Knowledge Agent Benchmark) is an agentic benchmark published by MKA Benchmark Leaderboard. Google DeepMind's benchmark of professional knowledge work where the key information sits in visual files such as architectural drawings, engineering schematics and medical scans. The agent has to deliver a finished report, spreadsheet, slide deck or marked-up drawing. Artificial Analysis runs it on a private 200-task set. Its evaluation set is described by the source as 200 multimodal professional tasks across 10 industry verticals. The source rates its difficulty as expert multimodal knowledge work.
Each model's result is published by MKA Benchmark Leaderboard as climb WSR (weighted rubric satisfaction rate, %). ModelChoice prints that value to 1 decimal place, exactly as published, and never converts or rescales it. This page ranks the 22 models that carry a published value, highest value first; a model the source never scored is left out rather than ranked at zero. The source does not state whether a higher value is the better result, so neither does this page.
| 4 | 53.4 |
| 5 | Anthropic | 53.0 |
| 6 | OpenAI | 52.7 |
| 7 | Anthropic | 52.0 |
| 8 | 51.3 |
| 9 | OpenAI | 50.2 |
| 10 | OpenAI | 47.7 |
| 11 | Meta | 47.5 |
| 12 | Moonshot AI | 43.9 |
| 13 | OpenAI | 43.8 |
| 14 | Anthropic | 41.0 |
| 15 | xAI | 40.1 |
| 16 | Xiaomi | 37.3 |
| 17 | OpenAI | 36.8 |
| 18 | Z.AI | 36.2 |
| 19 | 35.3 |
| 20 | Mistral | 35.1 |
| 21 | MiniMax | 33.5 |
| 22 | StepFun | 27.6 |
Evidence key: Observed
Rows are ordered by the value MKA Benchmark Leaderboard published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.