Startrise Frontend
Twelve single-shot frontend briefs, built in a real browser and scored on craft, engineering, brief adherence and originality. The blend is 25% deterministic gates, 45% a blind cross-lab judge panel and 30% a blind human review.
As of 21 Oct 2026, Claude Opus 5 leads Startrise Frontend on BenchLeader with 82.3%, ahead of Claude Opus 5.5 at 80.2%, across 29 model configurations with a published result.
- Published by
- Startrise
- Category
- Coding
- Index weight
- Reference only
- Models
- 29
- Data as of
- 21 Oct 2026
MIT — startriseio/sr-llm-benchmark; scores published as JSON per run.
- 1Claude Opus 582.3%
- 2Claude Opus 5.5 (high)80.2%
- 3GPT-6 Astra79.9%
- 4Kimi K379.8%
- 5GPT-6.1 Sol78.6%
- 6Claude Fable 5.177.1%
- 7Grok 4.775.9%
- 8Claude Sonnet 5.575.9%
- 9DeepSeek V4.1 Flash75.6%
- 10Muse Spark 1.374.7%
- 11Claude Fable 573.8%
- 12GLM 5.373.2%
- 13Claude Sonnet 5.5 (high)72.8%
- 14Grok 4.672.1%
- 15GLM 5.272.1%
29 of 29
| # | ||||
|---|---|---|---|---|
| 1 |  Claude Opus 5 | 82.3% | 65.4 | 2026-07-26 |
| 2 | Claude Opus 5.5high | 80.2% | 69.4 | 2026-10-05 |
| 3 |  GPT-6 Astra | 79.9% | 67.9 | 2026-10-05 |
| 4 |  Kimi K3open ↗ | 79.8% | 59.1 | 2026-07-26 |
| 5 | GPT-6.1 Sol | 78.6% | 63.3 | 2026-10-05 |
| 6 | Claude Fable 5.1 | 77.1% | 66.2 | 2026-10-05 |
| 7 |  Grok 4.7 | 75.9% | – | 2026-10-05 |
| 8 | Claude Sonnet 5.5 | 75.9% | – | 2026-10-05 |
| 9 |  DeepSeek V4.1 Flashopen ↗ | 75.6% | – | 2026-10-05 |
| 10 |  Muse Spark 1.3 | 74.7% | 58.3 | 2026-10-05 |
| 11 | Claude Fable 5 | 73.8% | 67.1 | 2026-07-26 |
| 12 |  GLM 5.3open ↗ | 73.2% | – | 2026-10-05 |
| 13 | Claude Sonnet 5.5high | 72.8% | 65.7 | 2026-10-05 |
| 14 | Grok 4.6 | 72.1% | 57.8 | 2026-08-12 |
| 15 | GLM 5.2open ↗ | 72.1% | 52.0 | 2026-07-26 |
| 16 |  Qwen3.8 Maxmax | 71.3% | 63.3 | 2026-08-11 |
| 17 | Claude Opus 5.5 | 71.1% | 61.9 | 2026-10-05 |
| 18 | Muse Spark 1.2 | 70.3% | – | 2026-08-10 |
| 19 | Qwen3.8 Max | 69.8% | – | 2026-10-05 |
| 20 | GPT-6 Luna | 69.3% | – | 2026-10-05 |
| 21 | Grok 4.5 | 68.0% | 58.9 | 2026-07-26 |
| 22 | GPT-5.6 Terra | 66.9% | 52.6 | 2026-07-26 |
| 23 | Claude Sonnet 5 | 66.6% | 56.0 | 2026-07-26 |
| 24 | GPT-5.6 Sol | 66.2% | 55.9 | 2026-07-26 |
| 25 | GPT-5.6 Luna | 65.0% | 47.9 | 2026-07-26 |
| 26 |  Gemini 3.8 Flash | 64.6% | 55.6 | 2026-10-05 |
| 27 | Qwen3 7max | 63.6% | 60.7 | 2026-07-26 |
| 28 | DeepSeek V4 Proopen ↗ | 62.7% | 53.5 | 2026-10-05 |
| 29 | Claude Haiku 4.5 | 44.8% | 45.3 | 2026-07-26 |
More coding benchmarks · All LLM benchmarks
Cite as: BenchLeader, “Startrise Frontend leaderboard”, https://www.benchleader.com/benchmarks/startrise_frontend, data as of 21 Oct 2026.
Startrise Frontend: questions
- What does Startrise Frontend measure?
- Twelve single-shot frontend briefs, built in a real browser and scored on craft, engineering, brief adherence and originality. The blend is 25% deterministic gates, 45% a blind cross-lab judge panel and 30% a blind human review. Scores are reported in percent of tasks solved; higher is better.
- Which AI model leads Startrise Frontend?
- Claude Opus 5 leads Startrise Frontend with 82.3% as of 21 Oct 2026, ahead of Claude Opus 5.5 at 80.2%.
- How many models have Startrise Frontend results?
- 29 model configurations have a Startrise Frontend result on BenchLeader, all taken from Startrise.
- Who runs Startrise Frontend and how often is it updated?
- Startrise Frontend is published by Startrise. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does Startrise Frontend count toward the BenchLeader Index?
- No. Startrise Frontend is shown for reference but left out of the composite index.