Tool
Which coding model scores highest for the least money?
Price basis: Blended (3 input : 1 output)
Head-to-head Elo rating for front-end web development, from the LMArena WebDev leaderboard.
WebDev Arena vs API price per 1M blended tokens · up and to the left is better
Scores updated 20 Oct 2026 · snapshot
Each dot is one model. Up and to the left is better: higher score, lower price. The teal staircase is the answer to "what is the best model I can get at a given price?" — at any point on the price axis, the model at the top of that step is the best you can buy for that much or less. Everything below the line is beaten by something cheaper.
Head-to-head Elo rating for front-end web development, from the LMArena WebDev leaderboard.
Notes on this data
Benchmark scores show performance on test tasks only — not your real day-to-day coding experience. Where a source lists several reasoning-effort settings, we show the model's best listed result. Claude, Cursor and ChatGPT plans show the monthly fee only: their usage allowances are not officially published per model.
Models16 of 31 selectedShowHide
- Anthropic
- Anthropic
- Anthropic
- Anthropic
- OpenAI
- OpenAI
- OpenAI
API price not yet in our verified price list
- OpenAI
- OpenAI
- OpenAI
API price not yet in our verified price list
- Google
- Google
- Google
- Google
- Google
- DeepSeek
- DeepSeek
- Moonshot (Kimi)
- Moonshot (Kimi)
- Moonshot (Kimi)
API price not yet in our verified price list
- Z.ai
- Z.ai
- Z.ai
- Alibaba (Qwen)
- Alibaba (Qwen)
- Alibaba (Qwen)
No WebDev Arena score published
- MiniMax
- MiniMax
- xAI
- xAI
API price not yet in our verified price list
- Cognition
No WebDev Arena score published
Selection is saved in this browser. Click a model name to see where it is available.
Add a custom point
For the WebDev Arena score and the current API price basis only. Saved in this browser, never sent anywhere, and kept out of the value frontier and the answer.
Data table · 16 plotted points
| Model | Vendor | WebDev Arena | API price | Value frontier | Cheapest paid plan |
|---|---|---|---|---|---|
| Anthropic | 1586High | $0.20/1M blended tokens | Yes | OpenCode Go | |
| OpenAI | 1578Max | $0.20/1M blended tokens | No | ChatGPT Go | |
| DeepSeek | 1664Max | $0.52/1M blended tokens | Yes | Command Code Go | |
| MiniMax | 1482default | $0.52/1M blended tokens | No | ClinePass | |
| 1585High | $1.50/1M blended tokens | No | Not on a paid plan | ||
| DeepSeek | 1582High, 20260813 | $1.98/1M blended tokens | No | QwenCloud Token Plan Lite | |
| Z.ai | 1621Max | $2.15/1M blended tokens | No | ClinePass | |
| xAI | 1637xHigh | $3/1M blended tokens | No | Copilot Pro+ | |
| Alibaba (Qwen) | 16740902 | $3/1M blended tokens | Yes | QwenCloud Token Plan Lite | |
| Anthropic | 1768xHigh | $4/1M blended tokens | Yes | Claude Pro | |
| OpenAI | 1753Max | $4/1M blended tokens | No | ChatGPT Plus | |
| Moonshot (Kimi) | 1655Max | $6/1M blended tokens | No | ClinePass | |
| Anthropic | 1807Max | $8/1M blended tokens | Yes | Claude Pro | |
| 1678High | $8/1M blended tokens | No | OpenCode Go | ||
| Anthropic | 1743Max | $20/1M blended tokens | No | Claude Pro | |
| OpenAI | 1774Max | $20/1M blended tokens | No | ChatGPT Plus |
Data sources
Scores from the committed snapshot.
Benchmarks
Epoch AI, 'Capabilities & benchmarking'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks'. epoch.ai/benchmarks · Licence CC BY 4.0.
FrontierCode (Cognition), FrontierSWE, DeepSWE (DataCurve) and CursorBench (Cursor) are external benchmarks compiled by Epoch AI. Links to each original source: FrontierCode (Cognition), FrontierSWE (FrontierSWE), DeepSWE (DataCurve), CursorBench (Cursor). Benchmark questions belong to their creators.
WebDev Arena scores from the LMArena leaderboard dataset on Hugging Face (lmarena-ai/leaderboard-dataset), CC BY 4.0. Dataset · leaderboard published 19 Oct 2026. Licence CC BY 4.0.
Prices
API prices and plan prices: BestAIPlan’s own source-checked data. API list updated 21 Oct 2026 · methodology.
Dates
- Scores generated 20 Oct 2026 (snapshot)
- Epoch AI data fetched 20 Oct 2026
- LMArena leaderboard published 19 Oct 2026
Refresh process
Scores refresh daily. Large jumps are held for manual review, and a new model appears once we register its name mapping — usually within 1–2 days of release.