SWE Atlas: Test Writing leaderboard
As of 6 Nov 2026, the top score on SWE Atlas: Test Writing is 67.0% by Claude Fable 5.1 (Anthropic), ahead of Claude Opus 5 at 62.2%, across 22 model configurations with a published result (Scale AI SEAL, result dated 2 Sep 2026). The model writes tests for real code and is scored on coverage and correctness. SWE Atlas: Test Writing is shown for reference only and does not count toward the BenchLeader Index.
Key facts (as of 6 Nov 2026)
- Claude Fable 5.1 holds the top SWE Atlas: Test Writing score at 67.0% (Scale AI SEAL, 2 Sep 2026).
- Claude Opus 5 is second on SWE Atlas: Test Writing at 62.2% (Scale AI SEAL, 4 Aug 2026).
- SWE Atlas: Test Writing: Writing tests for existing code. Scale AI (Scale AI SEAL).
- 22 model configurations have a published SWE Atlas: Test Writing result on BenchLeader (BenchLeader, 6 Nov 2026).
- SWE Atlas: Test Writing does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).
Writing tests for existing code. Scale AI.
- Published by
- Scale AI SEAL
- Category
- Coding
- Index weight
- Reference only
- Models
- 22
- Data as of
- 6 Nov 2026
Scale AI SEAL Leaderboards.
What the test looks like
The model writes tests for real code and is scored on coverage and correctness.
How it is scored
Score, published by Scale AI.
What to keep in mind
A specific engineering skill.
- 1Claude Fable 5.1 (xhigh)67.0%
- 2Claude Opus 5 (xhigh)62.2%
- 3Claude Fable 5 (xhigh)55.6%
- 4Gemini 3.8 Flash53.7%
- 5GPT-6 Astra (xhigh)50.7%
- 6Claude Opus 4.8 (xhigh)49.6%
- 7GPT-5.6 Sol (xhigh)45.9%
- 8GPT-5.4 (xhigh)44.4%
- 9GPT-5.5 (xhigh)42.6%
- 10Muse Spark 1.1 (xhigh)41.5%
- 11GLM 5.241.5%
- 12GPT-5.3 Chat (xhigh)39.0%
- 13Claude Opus 4.738.5%
- 14Claude Opus 4.636.7%
- 15Claude Sonnet 4.631.8%
| # | ||||
|---|---|---|---|---|
| 1 |  Claude Fable 5.1xhigh | 67.0% | 65.4 | 2026-09-02 |
| 2 | Claude Opus 5xhigh | 62.2% | 60.2 | 2026-08-04 |
| 3 | Claude Fable 5xhigh | 55.6% | – | 2026-06-11 |
| 4 |  Gemini 3.8 Flash | 53.7% | 54.4 | 2026-09-09 |
| 5 |  GPT-6 Astraxhigh | 50.7% | 66.8 | 2026-09-09 |
| 6 | Claude Opus 4.8xhigh | 49.6% | – | 2026-06-08 |
| 7 | GPT-5.6 Solxhigh | 45.9% | 64.2 | 2026-07-28 |
| 8 | GPT-5.4xhigh | 44.4% | 63.2 | 2026-03-26 |
| 9 | GPT-5.5xhigh | 42.6% | 64.3 | 2026-05-07 |
| 10 |  Muse Spark 1.1xhigh | 41.5% | 59.5 | 2026-07-09 |
| 11 |  GLM 5.2open ↗ | 41.5% | 51.5 | 2026-06-23 |
| 12 | GPT-5.3 Chatxhigh | 39.0% | – | 2026-03-26 |
| 13 | Claude Opus 4.7 | 38.5% | 63.1 | 2026-06-18 |
| 14 | Claude Opus 4.6 | 36.7% | 62.3 | 2026-03-26 |
| 15 | Claude Sonnet 4.6 | 31.8% | 58.2 | 2026-03-26 |
| 16 | Muse Spark | 31.1% | 64.1 | 2026-04-08 |
| 17 | Gemini 3 Flash | 30.3% | 56.9 | 2026-03-26 |
| 18 | Gemini 3.1 Pro | 29.8% | 60.5 | 2026-03-26 |
| 19 | GLM-5open ↗ | 28.7% | 54.8 | 2026-03-26 |
| 20 |  DeepSeek V4 Proopen ↗ | 27.1% | 53.2 | 2026-06-18 |
| 21 |  Kimi K2.5open ↗ | 25.8% | 54.2 | 2026-03-26 |
| 22 |  MiniMax M2.5open ↗ | 18.6% | 53.6 | 2026-03-26 |
More coding benchmarks · All LLM benchmarks
Related benchmarks
Top head-to-heads on SWE Atlas: Test Writing
Cite as: BenchLeader, “SWE Atlas: Test Writing leaderboard”, data as of 6 Nov 2026, https://www.benchleader.com/benchmarks/scale_sweatlas_tw.
SWE Atlas: Test Writing: questions
- What does SWE Atlas: Test Writing measure?
- The model writes tests for real code and is scored on coverage and correctness. Scores are reported in percent of tasks solved; higher is better.
- What is the top score on SWE Atlas: Test Writing?
- As of 6 Nov 2026, the highest score on SWE Atlas: Test Writing is 67.0% by Claude Fable 5.1 (result dated 2 Sep 2026), ahead of Claude Opus 5 at 62.2%. Scores come from Scale AI SEAL and are re-read daily.
- Which AI model leads SWE Atlas: Test Writing?
- Claude Fable 5.1 leads SWE Atlas: Test Writing with 67.0% as of 6 Nov 2026 (result dated 2 Sep 2026), ahead of Claude Opus 5 at 62.2%.
- How many models have SWE Atlas: Test Writing results?
- 22 model configurations have a SWE Atlas: Test Writing result on BenchLeader, all taken from Scale AI SEAL.
- Who runs SWE Atlas: Test Writing and how often is it updated?
- SWE Atlas: Test Writing is published by Scale AI SEAL. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does SWE Atlas: Test Writing count toward the BenchLeader Index?
- No. SWE Atlas: Test Writing is shown for reference but left out of the composite index.