Skip to content

Artificial Analysis Review

Artificial Analysis is an independent AI benchmarking site that ranks language, image, video, speech and music models and the APIs that serve them on intelligence, speed and price. Free to browse; Pro is $499 a month.

Free plan · from $499/mo · Web

Stork's verdict on Artificial Analysis

Artificial Analysis is the most useful free place to compare models and API providers on intelligence, speed and price, but Pro is priced for companies, not individuals.

Artificial Analysis reviewed by Stork AI · stork.ai/en/artificial-analysis

Two calls. That is the integration.
A talking-head API at 20 cents a minute.
Send a photo and text or audio, get an MP4. Prepaid credits. Failed jobs are refunded in full.
Read the docsNo code? Make a photo talk

What Artificial Analysis does

Artificial Analysis (artificialanalysis.ai) runs its own benchmarks on AI models, the cloud APIs that host them and the chips they run on, and publishes the results as public leaderboards and charts. Its About page lists 500+ models benchmarked, 100+ inference providers and 1,000+ endpoints. The headline number is the Artificial Analysis Intelligence Index, a composite score. Version 5, announced October 27, 2026, weights knowledge work, coding, science and general knowledge at 25% each, adds Cognition's FrontierCode 2.0 and Terminal-Bench-Science, and says 60% of the index now rests on private test sets or held-out answers. Beyond language models it covers coding agents, text-to-image, video, text-to-speech, speech-to-text and music, including community arenas for image, video and speech.

Performance benchmarks measure what customers see end to end (output speed, time to first token, end-to-end time per task) rather than the peak a chip can reach, and prices are tracked per endpoint. The company says providers cannot pay for results, methodology changes or listing. Signed-in tools include a Model Recommender, MicroEvals for comparing models on your own prompts, and Optima for turning your own tasks into a repeatable benchmark that compares quality against cost per task. Data is also available through a REST API and an MCP server.

Artificial Analysis pricing

The public website, leaderboards and a free API tier cost nothing. Paid plans add data tools, exports, fuller API access and reports.

PlanMonthly priceIncluded
Free$0Public leaderboards and charts; free API key with the language model list (headline indices, median performance, input and output prices) and free media endpoints, 100 requests per 24 hours (200 signed-in MCP calls)
Pro$499/monthPer seat. Data Playground, Table Builder, Data Export, Optima, language and media model data via API and databook download, State of AI Report and AI Adoption Survey, email support, 500 API requests per 24 hours (1,000 signed-in MCP calls)
EnterpriseCustomEverything in Pro plus language model provider data, highest API rate limits, compute market model with forecasts, Model Deployment Report, dedicated support, workshops, AI advisory and custom benchmarking; aimed at organizations with 150+ employees

Yearly billing brings Pro to $417 a month per seat, which the page says saves $984 a year ($499 x 12 = $5,988 vs $417 x 12 = $5,004). Pro covers a single seat; more seats need Enterprise. There is no free trial. Cards are accepted, and Enterprise can pay by invoice; academic discounts are available on request. The API docs also list a Commercial tier (Pro plus provider-level data, performance over time and raw measurements) with custom rate limits, priced by contacting sales. Every API tier requires visible attribution to Artificial Analysis, and redistribution or external use of the data needs a separate agreement.

Prices checked on the vendor's pricing page on November 10, 2026. See Artificial Analysis pricing and the Data API docs.

Key features

What Artificial Analysis offers, from its own site and docs:

  • Intelligence Index v5 composite score plus separate indexes such as Coding Agents, Cyber, Decision and Openness
  • Language model leaderboard comparing intelligence, price, output speed, time to first token and context window
  • Provider-by-provider benchmarks of the same model across serverless APIs, measured end to end
  • Image, video and text-to-speech arenas and leaderboards, plus speech-to-text and music coverage
  • Model Recommender that ranks models by your priorities for intelligence, speed and price
  • MicroEvals for side-by-side comparisons on your own prompts, and Optima for custom task benchmarks with cost per task (Pro)
  • REST API (31 endpoints) and an MCP server at mcp.artificialanalysis.ai, with free, Pro and Commercial tiers
  • Published methodology pages for each benchmark and a changelog

Who Artificial Analysis is best for

The free site suits developers and product teams picking a model or an API provider: you can compare intelligence scores against price and speed for the same model across hosts without signing up. Pro is priced for companies that want to export the data, build custom tables, run Optima on their own tasks or pull more than the free API allows; at $499 a month per seat it is a business tool rather than a hobbyist one.

If you want rankings based on human preference votes rather than fixed benchmarks, look at LMArena. If you want rankings by what developers actually send through an API, OpenRouter's rankings use token volume. For academic, transparent evaluation across many scenarios, Stanford's HELM is free and open.

What users say

Artificial Analysis is widely cited by AI labs and press, and its Intelligence Index is often quoted in model launches. DeepLearning.AI's The Batch welcomed the January 2026 move from saturated tests to harder business tasks but said the suite still did not separate the top models much; the article discloses that Andrew Ng, who founded DeepLearning.AI, is an investor. Nathan Lambert of Interconnects argues that collapsing many capabilities into one rankable number is what makes the index useful and also what makes it misleading. A Latent Space writeup describes the team's "mystery shopper" accounts for testing labs anonymously; Latent Space co-founder swyx is listed among Artificial Analysis backers. No independent review sites with user ratings for the paid plan were found.

Pros

  • Free leaderboards compare intelligence, price and speed in one place
  • Benchmarks the same model across many API providers, not just the lab's own
  • Runs its own evaluations, with 60% of Index v5 on private or held-out sets
  • Covers image, video, speech and music models as well as LLMs
  • Free API and MCP server for pulling headline data into your own tools

Cons

  • Pro costs $499 a month per seat with no free trial
  • Provider-level API data needs Enterprise or Commercial, priced on request
  • A single composite index can hide large differences on specific tasks
  • Index versions change benchmarks, so scores are not comparable across versions

Frequently asked questions

Is Artificial Analysis free?

The website, leaderboards and a free API tier are free. The free API returns the language model list with headline indices, median performance and prices, plus free media endpoints, with 100 requests per 24 hours. Pro costs $499 a month per seat, or $417 a month billed yearly.

Does Artificial Analysis offer a free trial of Pro?

No. The pricing page FAQ says it does not currently offer free trials. Academic institutions can ask about discounts.

What is the Artificial Analysis Intelligence Index?

A composite score of model capability. Version 5 (October 27, 2026) weights knowledge work, coding, science and general knowledge at 25% each, using benchmarks such as AA-Briefcase, FrontierCode 2.0, Terminal-Bench 4.0, Terminal-Bench-Science, CritPt v2, AA-Omniscience v1.1 and HLE-Diamond. 60% of it is based on private or held-out test sets.

Can AI companies pay to rank higher on Artificial Analysis?

The company says no: its independence policy states providers cannot pay for results, methodology changes or listing on the website. It does sell reports, custom benchmarking and advisory to enterprises.

Does Artificial Analysis have an API?

Yes. A REST API at artificialanalysis.ai/api/v2 and an MCP server. Free keys get 100 requests per 24 hours, Pro 500, Commercial custom. Signed-in MCP use gets double. All tiers require attribution when you display the data.

How does Artificial Analysis measure speed?

It benchmarks end-to-end performance as a customer experiences it, such as output tokens per second, time to first token and time per task, across providers. Results are meant to reflect real-world API performance, not the maximum possible on given hardware.

Who runs Artificial Analysis?

Its About page names co-founders Micah Hill-Smith (CEO) and George Cameron (CPO), based in San Francisco, and lists backers including Nat Friedman, Daniel Gross and Andrew Ng.

Compare Alternatives

Artificial Analysis

Free plan · from $499/mo · Web

From the Stork blog

Enjoying this? Get one like it in your inbox each morning.

one email a day · unsubscribe in two clicks · no third-party tracking