FrontierSWE leaderboard
As of 6 Nov 2026, the top score on FrontierSWE is 65.5% by GPT-6 Astra (OpenAI), ahead of Claude Opus 5.5 at 62.3%, across 20 model configurations with a published result (Epoch AI Benchmarking Hub, result dated 3 Sep 2026). Long software tasks split into implementation, performance and research work, run in an agent harness. FrontierSWE is shown for reference only and does not count toward the BenchLeader Index.
Key facts (as of 6 Nov 2026)
- GPT-6 Astra holds the top FrontierSWE score at 65.5% (Epoch AI Benchmarking Hub, 3 Sep 2026).
- Claude Opus 5.5 is second on FrontierSWE at 62.3% (Epoch AI Benchmarking Hub, 22 Sep 2026).
- FrontierSWE: Frontier-difficulty software-engineering tasks across implementation, performance and research (Epoch AI Benchmarking Hub).
- 20 model configurations have a published FrontierSWE result on BenchLeader (BenchLeader, 6 Nov 2026).
- FrontierSWE does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).
Frontier-difficulty software-engineering tasks across implementation, performance and research.
- Published by
- Epoch AI Benchmarking Hub
- Category
- Coding
- Index weight
- Reference only
- Models
- 20
- Data as of
- 6 Nov 2026
CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.
What the test looks like
Long software tasks split into implementation, performance and research work, run in an agent harness.
How it is scored
Score, as published.
What to keep in mind
Very new; a dozen models.
- 1GPT-6 Astra (max)65.5%
- 2Claude Opus 5.5 (max)62.3%
- 3Claude Sonnet 5.5 (max)61.9%
- 4Claude Fable 5.1 (max)56.3%
- 5Gemini 4 Argon (high)55.0%
- 6Claude Opus 5 (max)52.0%
- 7Claude Fable 5 (max)47.0%
- 8Claude Haiku 5.5 (max)43.8%
- 9GPT-5.6 Sol (max)32.2%
- 10GLM 5.3 (max)30.2%
- 11Grok 4.7 (xhigh)29.5%
- 12Kimi K3 (max)25.9%
- 13Muse Spark 1.3 (max)25.8%
- 14Grok 4.6 (xhigh)25.3%
- 15Gemini 3.7 Flash (high)20.3%
| # | ||||
|---|---|---|---|---|
| 1 |  GPT-6 Astramax | 65.5% | 67.8 | 2026-09-03 |
| 2 |  Claude Opus 5.5max | 62.3% | 69.3 | 2026-09-22 |
| 3 | Claude Sonnet 5.5max | 61.9% | 65.4 | 2026-09-28 |
| 4 | Claude Fable 5.1max | 56.3% | 67.2 | 2026-09-01 |
| 5 |  Gemini 4 Argonhigh | 55.0% | 66.0 | 2026-09-30 |
| 6 | Claude Opus 5max | 52.0% | 65.4 | 2026-07-24 |
| 7 | Claude Fable 5max | 47.0% | 65.3 | 2026-06-09 |
| 8 | Claude Haiku 5.5max | 43.8% | 57.0 | 2026-10-07 |
| 9 | GPT-5.6 Solmax | 32.2% | 65.4 | 2026-07-09 |
| 10 |  GLM 5.3maxopen ↗ | 30.2% | 59.0 | 2026-08-14 |
| 11 |  Grok 4.7xhigh | 29.5% | 58.3 | 2026-09-21 |
| 12 |  Kimi K3maxopen ↗ | 25.9% | 62.1 | 2026-07-16 |
| 13 |  Muse Spark 1.3max | 25.8% | 62.0 | 2026-09-02 |
| 14 | Grok 4.6xhigh | 25.3% | 57.9 | 2026-08-12 |
| 15 | Gemini 3.7 Flashhigh | 20.3% | 62.2 | 2026-08-13 |
| 16 | Gemini 3.8 Flashhigh | 19.6% | 62.0 | 2026-09-02 |
| 17 | GLM 5.3 Flashmaxopen ↗ | 18.1% | 54.2 | 2026-08-20 |
| 18 |  Qwen3.8 Maxxhigh | 17.8% | 57.1 | 2026-09-01 |
| 19 | Muse Spark 1.2xhigh | 12.0% | 61.7 | 2026-08-05 |
| 20 | Inklingxhighopen ↗ | 4.1% | 50.2 | 2026-07-15 |
More coding benchmarks · All LLM benchmarks
Related benchmarks
Top head-to-heads on FrontierSWE
Cite as: BenchLeader, “FrontierSWE leaderboard”, data as of 6 Nov 2026, https://www.benchleader.com/benchmarks/frontierswe.
FrontierSWE: questions
- What does FrontierSWE measure?
- Long software tasks split into implementation, performance and research work, run in an agent harness. Scores are reported in percent of tasks solved; higher is better.
- What is the top score on FrontierSWE?
- As of 6 Nov 2026, the highest score on FrontierSWE is 65.5% by GPT-6 Astra (result dated 3 Sep 2026), ahead of Claude Opus 5.5 at 62.3%. Scores come from Epoch AI Benchmarking Hub and are re-read daily.
- Which AI model leads FrontierSWE?
- GPT-6 Astra leads FrontierSWE with 65.5% as of 6 Nov 2026 (result dated 3 Sep 2026), ahead of Claude Opus 5.5 at 62.3%.
- How many models have FrontierSWE results?
- 20 model configurations have a FrontierSWE result on BenchLeader, all taken from Epoch AI Benchmarking Hub.
- Who runs FrontierSWE and how often is it updated?
- FrontierSWE is published by Epoch AI Benchmarking Hub. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does FrontierSWE count toward the BenchLeader Index?
- No. FrontierSWE is shown for reference but left out of the composite index.