Model finder

Three questions, then the models nothing beats on score and price among those that meet them. It is the leaderboard's own ranking: must-haves decide which models compete, and the frontier is computed among those alone. It runs in your browser: changing an answer sends no request.

1 · What is it for?
2 · Budget
3 · Must have

Answer

12 undominated of 358 models on the leaderboard that meet your answers. 213 of those have no independent score, so they are listed apart and never ranked.

6 more models meet them but are sold only as a free listing. The leaderboard shows those under “Show batch & free”; they are not counted or ranked here.

Prices are the balanced workload (3 tokens in per 1 out), in USD per million tokens, as of Oct 6, 2026. Open this view on the leaderboard.

Undominated models that meet your answers, best score first
ModelScore$/M
Claude Opus 5.51511.7$8.00
Gemini 3.8 Flash1497.0$3.00
MiMo-V2.6-Pro1491.0$0.540
GLM 5.3 Flash1469.6$0.238
DeepSeek V4.1 Flash1462.5$0.182
MiMo-V2.6-Flash1456.4$0.175
Gemma 4 26B A4B 1433.9$0.127
DeepSeek V4 Flash 04231432.1$0.113
gpt-oss-120b1365.4$0.065
Gemma 3 4B1290.8$0.063
gpt-oss-20b1287.7$0.036
Llama 3.1 8B Instruct1186.5$0.025
The 213 on the leaderboard with no independent score, by name

General scores are LMArena Elo under CC BY 4.0, from the official dataset.

Evidence & Ask