Artificial Analysis (Batch 4)

Artificial Analysis publishes independent benchmarks of frontier AI models. The company was founded in 2024 by George Cameron and Micah Hill-Smith, both ex-Google interns, and is based in Sydney. The site at artificialanalysis.ai compares more than 100 models across providers on intelligence, speed, price, and context window.

The methodology page documents the metrics. Performance benchmarks measure end-to-end performance experienced by API customers rather than theoretical hardware maximums. Tracked figures include time to first token, time to first answer token for reasoning models, output speed in tokens per second, end-to-end response time, and average reasoning tokens across a fixed 60-prompt suite. Pricing is reported as separate input, output, and cache-hit prices per token, normalized using OpenAI's tokenizer for cross-provider comparison, with a blended price calculated at a 3:1 input-to-output ratio. The Intelligence Index, currently at v4.0, is a composite of ten evaluations including GDPval-AA, tau-squared-Bench Telecom, and Terminal-Bench Hard.

Revenue comes from a Sydney-based team of roughly 20 building two complementary products: the public leaderboards and a paid private benchmarking service for enterprises and AI labs that want bespoke evaluation against internal specs. The company publishes a quarterly State of AI Highlights Report; Jensen Huang cited the firm in his Nvidia CES 2026 keynote, and the team are repeat guests on the Latent Space podcast.

AI Grant Batch 4 selected Artificial Analysis for a $250,000 SAFE plus Microsoft Azure and partner credits, the standard Batch 4 structure run by Nat Friedman and Daniel Gross. Subsequent reporting notes that Artificial Analysis was among the few Batch 4 companies to also raise a full seed round from Friedman and Gross directly. The grant places the company at the evaluation meta-layer, supplying a third-party reference for buyers comparing closed and open model APIs.