Model catalogue
You get access to the AI models from the major providers through one subscription. New models enter the catalogue as soon as they are released and we have assessed them.
The AI models
The models are grouped by provider. Where an older version is still the better choice for some work, because it is cheaper or faster, it stays in the list alongside the newer one.
- Anthropic
- Claude Opus 5
- Claude Sonnet 5
- Claude Haiku 4.5
- OpenAI
- GPT-5.6 Sol
- GPT-5.6 Terra
- GPT-5.6 Luna
- xAI
- Grok 4.6
- Grok 4.5
- Moonshot AI
- Kimi K3
- Kimi K2.6
- Google
- Gemini 3.8 Flash
- Gemini 3.5 Flash-Lite
- Alibaba
- Qwen3.8 Max
- Qwen3.7 Plus
- Mistral AI
- GLM 5.2
- Mistral Medium
- Mistral Small
How we choose models
The catalogue is not a ranking. It is meant to give you a real choice: several providers, a small fast model beside a large careful one, a cheap one beside an expensive one, without making you read through every version each provider has ever released. Some models are in it precisely because they are small and cheap, not because they score highly.
Every model is measured the same way, on figures published by Artificial Analysis rather than on the providers’ own: how well it works, how fast it answers, and what it costs.
Quality
How well the model solves the tasks. The higher the score, the harder the problems it can handle, though most everyday work does not need the highest score in the catalogue.
Speed
How long the model takes to answer. Some tasks are worth waiting for a model that thinks them through, and others just need an answer straight away. The catalogue holds both, so you can pick the model the task calls for.
Price
What it costs to get the same work done. Prices differ a great deal between models, and the most expensive one is rarely the right choice for everyday tasks. The catalogue therefore lists both the powerful and the cheap ones, so you do not pay for more than the task needs.