Quality against price
- Frontier
- Other models
- Search match
A daily market sheet for language models: what each point of quality costs.
Each model's ELO score, with style control, comes from the official LM Arena leaderboard dataset. Like a chess rating, it is built from blind votes between pairs of models: a 100-point lead means winning about 64% of head-to-head votes.
The price per million tokens blends the input and output rates listed by the OpenRouter models API, weighted by the token mix above. A ranked model gets a price only from an entry naming the same model and version. Use 1:0 to compare input prices alone.
A model is on the frontier when no other model is both cheaper and better ranked. The stepped line gives the best score you can buy at each price; the gap column says how far below that line a model sits.