Documentation
Rankings
Compare current AI models on their published benchmark scores and API providers on price, with every figure linked to where it was published.
Overview
Rankings covers 200+ current models, including Claude, GPT, Gemini, Grok, DeepSeek and Qwen. Open it at qai.io/bench/rankings and switch between two views:
| View | What it shows |
|---|---|
| LLMs | Benchmarks: each model’s published scores on reasoning, agentic and coding benchmarks. Pricing: each maker’s own API price, context window and licence. |
| LLM APIs | Benchmarks: each model’s scores with the providers that serve it. Pricing: each provider’s list price and the context window it serves. |
Reading the scores
Each column follows one setup of its benchmark, named in its header, so scores in a column compare like with like. A score its maker published, used where no other is available, is marked. Prices are each provider’s standard list price, and the page shows the date its data is as of.
Pick models and benchmarks for the comparison chart, filter the table by model or maker, and open any model for its prices and providers on Model pricing.
FAQ
Did QAI run these benchmarks?
No. Rankings gathers published results and links each one to its source. For results we measure ourselves, see EdgeBench.
What does “—” mean?
There is no score for that model in that column’s setup.
Is a provider’s score a test of its service?
No. The scores belong to the model. The LLM APIs view shows which providers serve it and what they charge.
Do I need an account?
No. Rankings is open to everyone.