Rankings
Loading the rankings
Loading.
Rankings
Every model, ranked
334 models, sorted by what the catalog publishes and what we measure. A typical message is about 500 tokens you send and 500 it writes.
Cheapest per 1,000 messages
Paid models by what 1,000 typical messages cost at the published price.
- 1Mistral NemoMistral3.2¢
- 2Ling 3.0 TinyinclusionAI3.3¢
- 3Llama 3.1 8B InstructMeta3.3¢
- 4Ling 3.0 FlashinclusionAI4.6¢
- 5Llama 3 8B LunarisSao10K4.9¢
- 6Qwen3.5-9BQwen6.2¢
- 7GLM 5.3 FlashZ.ai6.5¢
- 8Nex-N2.5-MiniNex AGI6.9¢
Biggest context
How many tokens a model can read at once, including the thread, your files and its answer.
- 1Grok 4.20xAI2M
- 2Grok 4.20 Multi-AgentxAI2M
- 3MiMo-V2.6-FlashXiaomi1M
- 4GPT-6 LunaOpenAI1M
Newest arrivals
By the release date the catalog publishes.
- 1Gemini 4 HydrogenGoogleNov 12, 2026
- 2ERNIE-Light-DT-1.0BaiduNov 10, 2026
- 3Gemini 4 CarbonGoogleNov 10, 2026
- 4Nex-N2.6Nex AGINov 10, 2026
We could not load performance measurements. Try again.
16 models are free to use and left off the price board. Find them on the map.
Most used
This shows what people use here. It does not measure quality. Accounts appear only as coarse buckets, never as addresses.
We could not load usage rankings. Try again.
Seasons
A season is a contest with fixed dates. Its board counts what people did in that time, and a closed season keeps its final standing. Only people with a public page are named.
We could not load the boards. Try again shortly.