FrontierSWE
Integrity rank #15 of 65 · 21 models scored · top score 65.51 · GPT-6 Astra
unknown
Strongest on discrimination, weakest on contamination resistance. Caveat: the public test set does not rule out training-data contamination.
Reported scores — protocols may differ. How we compare
| # | Model | Score | Evidence | Measured |
|---|---|---|---|---|
| 1 | GPT-6 Astra | 65.51 | unverifiedT1 | 2026-09-03 |
| 2 | Claude Opus 5.5 | 62.33 | unverifiedT1 | 2026-09-22 |
| 3 | Claude Sonnet 5.5 | 61.89 | unverifiedT1 | 2026-09-28 |
| 4 | Claude Fable 5.1 | 56.29 | unverifiedT1 | 2026-09-01 |
| 5 | Gemini 4 Argon | 54.96 | unverifiedT1 | 2026-09-30 |
| 6 | Claude Opus 5 | 52.01 | unverifiedT1 | 2026-07-24 |
| 7 | Claude Fable 5 | 46.96 | unverifiedT1 | 2026-06-09 |
| 8 | Claude Haiku 5.5 | 43.77 | unverifiedT1 | 2026-10-07 |
| 9 | GPT-5.6 Sol | 32.2 | unverifiedT1 | 2026-07-09 |
| 10 | GLM-5.3 | 30.18 | unverifiedT1 | 2026-08-14 |
| 11 | Grok 4.7 | 29.5 | unverifiedT1 | 2026-09-21 |
| 12 | Kimi K3 | 25.87 | unverifiedT1 | 2026-07-16 |
| 13 | Muse Spark 1.3 | 25.8 | unverifiedT1 | 2026-09-02 |
| 14 | Grok 4.6 | 25.29 | unverifiedT1 | 2026-08-12 |
| 15 | Gemini 3.7 Flash | 20.26 | unverifiedT1 | 2026-08-13 |
| 16 | Gemini 3.8 Flash | 19.63 | unverifiedT1 | 2026-09-02 |
| 17 | GLM-5.3-Flash | 18.13 | unverifiedT1 | 2026-08-20 |
| 18 | Qwen3.8 Max (0902) | 17.81 | unverifiedT1 | 2026-09-01 |
| 19 | Qwen 3.8 Max | 15.81 | unverifiedT1 | 2026-08-02 |
| 20 | Muse Spark 1.2 | 11.97 | unverifiedT1 | 2026-08-05 |
| 21 | Inkling | 4.12 | unverifiedT1 | 2026-07-15 |
Continue from this leaderboard into the tightest head-to-head reads.
Weighted composite. Each dimension opens to the source-backed sub-signals it is computed from.
Claims drawn from cited facts, not live model generation.
The recorded metadata does not specify what this benchmark measures.
1 cited fact
This benchmark has scores from 21 tracked models, with a recorded score range spanning 61.39.
2 cited facts
Its Benchmark Integrity Index score is 95, with grade A and rank 15 out of 65 tracked benchmarks.
4 cited facts
It is not flagged as saturated. The top recorded score is 65.51, with 34.49 points of headroom. The computed top-score cluster contains 1 models.
4 cited facts
The harness classification is consistent harness. Test-set privacy is recorded as unknown. The recorded contamination-window ratio is 0.286.
3 cited facts
FrontierSWE: FrontierSWE as reported in Epoch AI's Capabilities Index CSV.
GPT-6 Astra leads FrontierSWE at 65.51. The full leaderboard above lists every recorded measurement, not just the headline number.
21 models have recorded scores on FrontierSWE, spanning a score spread of 61.39.
tensor.news grades FrontierSWE A for integrity (score 95/100), ranking #15 of 65 benchmarks we assess across discrimination, saturation, contamination resistance, harness comparability, and freshness.
No — FrontierSWE still has headroom and continues to discriminate between models rather than bunching them at the ceiling.
Its test set is unknown, so contamination risk is on the table: a high score may partly reflect training-data overlap rather than capability.
Scores are reported under a consistent harness, so comparisons on FrontierSWE are reasonably apples-to-apples.
Every FrontierSWE measurement is source-backed and tagged reproduced, self-reported, or unverified; the underlying record cites epoch.ai/data/processed_data_for_eci.csv.