FrontierCS 2 - SQL Counterexample Search: leaderboard
FrontierCS 2 selected formal runs on 48 research engineering tasks, with each source-native coding harness. Paper v1 Table 3 fixes the effort settings; the subsequently added DeepSeek 1017 row leaves effort unspecified. Superseded attempts are not selected by score. Native bounded task scores are distinct from human-reference margins and from pass/fail verdicts. Explicit failed-run zeroes remain zero; unreported scores remain missing. Metric: Task score (0–100).
Metric: Task score (0–100). Source: frontier-cs.org. Saturation forecast: Estimated already saturated. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GLM-5.3 (Max) | 97.66 | #39 (GLM-5.3) |
| 2 | DeepSeek V4.1 Flash (1017) | 97.66 | #27 |
| 3 | Gemini 4 Argon (High) | 96.57 | #8 (Gemini 4 Argon) |
| 4 | DeepSeek V4.1 Flash (Max) | 95.93 | #35 (DeepSeek V4.1 Flash) |
| 5 | GLM-5.3 Flash (Max) | 95.93 | #72 (GLM-5.3 Flash) |
| 6 | Muse Spark 1.3 (High) | 95.93 | #29 (Muse Spark 1.3) |
| 7 | GPT-6 Astra (xHigh) | 95.84 | #6 (GPT-6 Astra) |
| 8 | Kimi K3 (High) | 95.84 | #33 (Kimi K3) |
| 9 | Claude Opus 5.5 (Max) | 95.55 | #5 (Claude Opus 5.5) |
| 10 | Muse Wonder (Max) | 95.44 | #15 (Muse Wonder) |
| 11 | GPT-6.1 Sol (xHigh) | 92.95 | #7 (GPT-6.1 Sol) |
No result here: #9 Claude Fable 5.1, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=frontiercs-2-sql-counterexample-search · How It Works · Data refreshed daily, snapshot 2026-10-26.