Artificial Analysis
2 Tools · 299 Intel
Artificial Analysis is an independent AI benchmarking organization, founded in 2023 by Micah Hill-Smith and George Cameron, that runs its own evaluations of AI models and inference providers rather than relying on self-reported figures from AI labs. Its free public leaderboards and Intelligence Index have become widely referenced industry benchmarks for comparing model quality, speed, and cost. The company is backed by investors including Nat Friedman, Daniel Gross, and Andrew Ng.
Is this you? Sign in with X to claim this profile.
Tools
Independent benchmarks comparing AI models and providers on intelligence, coding, speed, latency, and price.
Artificial Analysis's platform for creating and running custom benchmarks against AI models on your own tasks.
Intel
MiniMax-M3 scores 55 on the Artificial Analysis index with image and video input and a 1M context. Weights are promised within about ten days, and heavy abstention drives its low hallucination rate.
Artificial Analysis benchmarks OpenAI's built-in web search against 25 other search setups using one fixed model. It ranks 7th of 26 at 74, leads on factual accuracy, lags on multi-hop BrowseComp, and costs about $0.05 per task.
Artificial Analysis compares 14 models on MLCR-AA and HealthBench Professional. Anthropic models lead on completeness and raw score, OpenAI models on accuracy and concision, and length-adjusted scoring closes the gap, with GPT-6 Astra tying Sonnet 5.5 for first.
AutomationBench-AA scores agents on 657 SaaS workflow tasks, counting objectives met without guardrail violations. Top scores are about 49%, every model breaks rules, and Gemini 3.5 Flash matches GPT-5.5 at roughly 37% of the cost.
Artificial Analysis places Ideogram 4.5 at #23 in image editing and #34 in text-to-image. High quality costs $0.22 per edit and $0.10 per generation, lower tiers start at $0.008, and it is on the Ideogram API with open weights promised.
Mistral releases Large 4 weights (1.05T total, 52B active, 1M context). Artificial Analysis scores it 26 on its index, with 39% fewer output tokens than the Preview, 28% hallucination rate, halved pricing and a license that restricts hosted-service use.
Artificial Analysis finds Gemini 3.8 Flash TTS leads its Pronunciation Robustness benchmark at 89.5%, ahead of prior Gemini and SpaceXAI models, with particular strength on contextual disambiguation and shorthand expansion.
Artificial Analysis finds Gemini 3 Pro Preview first on its Intelligence Index, 3 points above GPT-5.1, leading five of ten evals including HLE and AA-Omniscience, with better token efficiency than rivals but high per-token cost.
Artificial Analysis launches a leaderboard for Cognition's private FrontierCode 1.1 benchmark, which grades coding agents on unit tests plus maintainer merge criteria such as diff size and scope. Claude Opus 5.5 leads at 52%, and nine models cluster between 43% and 52%.
Artificial Analysis runs Harvey's legal agent benchmark on 120 private tasks. Claude Fable 5 leads with a 14.2% all-pass rate, and many models fully pass none despite high per-criterion scores.
Artificial Analysis benchmarks Z AI's GLM-5.3 (753B total, 40B active, 1M context). It ties Kimi K3 at 60 on Index v4.1 and jumps on agentic work, but token use rose about 20%, cost per task is $0.68, and hallucination rate ticked up. Scores predate the v5 Index, where it scores 32.