Benchmark
Arena ELO Webdev - Image to Webdev
Coding benchmarks compare models on programming tasks. Higher scores are better.
All coding benchmarksScores from LMArena data.
- Models measured
- 59
- Scores recorded
- 69
- Sources
- 1
- IQ weight
- 2.53
Score distribution
- 1,540.7
- Mean
- 1,544
- Median
- 100.9
- Std dev
- 1,410.7
- 10th percentile
- 1,664
- 90th percentile
- 1,284.6
- Min
This benchmark has a weight of 2.53 for chat models, measured over 59 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 51 | MiniMax M3· LMArena data | 1,473.7 |
| 52 | Qwen3.6 Plus· LMArena data | 1,473 |
| 53 | Gemini 3 Flash· LMArena data | 1,451.3 |
| 54 | Gemini 3.5 Flash Lite· LMArena data | 1,449.3 |
| 55 | GPT-5.3 CodexCodex Harness· LMArena data | 1,443.7 |
| 56 | Gemini 3 Pro Preview· LMArena data | 1,440.9 |
| 57 | GPT 5.4· LMArena data | 1,440.8 |
| 58 | Kimi K2.5Thinking· LMArena data | 1,439.2 |
| 59 | Qwen3.5 397B A17B· LMArena data | 1,437.8 |
| 60 | Gemini 3 FlashMinimal· LMArena data | 1,430.9 |
| 61 | GPT 5.1High· LMArena data | 1,419.2 |
| 62 | Kimi K2.5Instant· LMArena data | 1,410.7 |
| 63 | Inkling Small· LMArena data | 1,409.9 |
| 64 | NVIDIA Nemotron 3.5 Super VL 120BNVFP4· LMArena data | 1,391.7 |
| 65 | Grok 4.3· LMArena data | 1,375.5 |
| 66 | Mistral Large 3· LMArena data | 1,361.4 |
| 67 | GPT 5.1· LMArena data | 1,348.3 |
| 68 | Gemini 3.1 Flash Lite· LMArena data | 1,338.1 |
| 69 | Gemini 2.5 Pro· LMArena data | 1,284.6 |
Showing 51–69 of 69 results.
Data credits: LMArena, CC BY 4.0 (licence)