AI Benchmark Digest — 2026-07-11
AI Benchmark Digest — 2026-07-11
Daily
New Scores From Top-10 Models (16)
- GPT-5.6 Sol on AutomationBench: 18.1 Pass Rate (%) (#1/10)
- GPT-5.6 Sol on Bullshit Benchmark: 27.3 BS Detection Rate (%) (#59/158)
- GPT-5.6 Sol on Chess Puzzles (Epoch AI): 64.0 Accuracy (%) (#1/55)
- GPT-5.6 Sol on Design Arena (Data Viz): 1341.0 Elo (#2/132)
- GPT-5.6 Sol on Design Arena (Website): 1346.0 Elo (#2/151)
- GPT-5.6 Sol on Epoch AI - Cursorbench: 67.2 Score (#4/28)
- GPT-5.6 Sol on Epoch AI - Scicode: 56.13 Score (#7/69)
- GPT-5.6 Sol on FrontierMath - Tier 4 (v2): 82.93 Accuracy (%, 41 private v2 problems) (#2/38)
- GPT-5.6 Sol on LisanBench: 0.0721 Mean Path Length / Current Maximum (#45/150)
- GPT-5.6 Sol on Mercor APEX: 40.0 Best Pass@1 (%) (#4/51)
- GPT-5.6 Sol on Mercor APEX: 40.0 Best Pass@1 (%) (#4/51)
- GPT-5.6 Sol on Ramp SWE-Bench: 82.28 Resolved (%) (#4/24)
- GPT-5.6 Sol on Senior SWE-Bench: 24.4 Tasteful Solve Rate (pass@1, %) (#3/10)
- GPT-5.6 Sol on You're Absolutely Right!: 2.875 Average anti-sycophancy score (1-5) (#8/21)
- Muse Spark 1.1 on Chatbot Arena (Code): 1540.0 Elo (#10/96)
- Muse Spark 1.1 on Chatbot Arena (Text): 1490.0 Elo (#6/374)
New #1 Leaders (9)
- RuneBench (Total Peak XP Rate (XP/min)): GPT-5.6 Sol (7400.0) beat GPT-5.5 (6370.0) by 1030.0.
- Benchmarks.bio - scBench-Long (Pass Rate (%)): GPT-5.6 Sol (42.5) beat Claude Opus 4.8 (28.57) by 13.93.
- Opus Magnum Bench (Human-normalized score (%)): GPT-5.6 Sol (xHigh) (70.86) beat Claude Fable 5 (High) (60.15) by 10.71.
- WolfBench (Average Score (%)): GPT-5.6 Sol (85.0) beat GPT-5.5 (79.0) by 6.0.
- Agents' Last Exam (Pass Rate (%)): GPT-5 Sol (30.6) beat Claude Opus 4.8 (27.0) by 3.6.
- Epoch AI - Critpt (Score): GPT-5.6 Sol (32.3) beat GPT-5.5 Pro (xHigh) (30.57) by 1.73.
- WeirdML (Average Score): GPT-5.6 Sol (High) (88.76) beat Claude Fable 5 (High) (87.85) by 0.91.
- GDP.pdf (Strict Pass Rate (%)): GPT-5.6 Sol (30.7) beat Claude Fable 5 (30.0) by 0.7.
- Epoch AI - ECI (ECI Score): GPT-5.6 Sol (161.61) beat GPT-5.5 Pro (xHigh) (161.01) by 0.6.
Don't miss what's next. Subscribe to Mikhail Doroshenko: