AI Benchmark Digest — 2026-07-10

View on AI Benchmark Hub

Daily

New Models (1)

• Muse Spark 1.1 — ELO 3315, #5/1465, above Claude Mythos Preview, below Gemini 3 Deep Think
  • SEAL - MultiNRC: 65.59 (#1/44)
  • SEAL - VisualToolBench (VTB): 44.77 (#1/21)
  • SEAL - SWE-Bench Pro (Public Dataset): 61.5 (#1/25)
  • SEAL - SWE-Bench Pro (Private Dataset): 51.5 (#1/14)
  • SEAL - Professional Reasoning Benchmark - Finance: 55.01 (#1/29)
  • SEAL - Professional Reasoning Benchmark - Legal: 57.05 (#1/29)
  • SEAL - MCP Atlas: 88.1 (#1/24)
  • SEAL - MultiChallenge: 75.3 (#2/30)
  • SEAL - Fortress: 12.4 (#54/56)

New Scores From Top-10 Models (25)

• Claude Fable 5 on KernelBench Hub - Hard: 24.0 Best % of Hardware Roofline (#4/13)
• Claude Fable 5 on KernelBench Hub - Mega: 18.7 Best Speedup vs Reference (x) (#2/11)
• Muse Spark 1.1 on SEAL - Fortress: 12.4 Score (#54/56)
• Muse Spark 1.1 on SEAL - MultiChallenge: 75.3 Score (#2/30)
• Muse Spark 1.1 on SvelteBench: 92.2 Average pass@1 (%) (#36/145)
• Muse Spark 1.1 on Vals AI (Vals Index): 68.41 Accuracy (%) (#5/35)
• Muse Spark 1.1 on Vals AI Code Migration: 31.11 Accuracy (%) (#11/27)
• Muse Spark 1.1 on Vals AI CorpFin v2: 71.29 Accuracy (%) (#2/121)
• Muse Spark 1.1 on Vals AI Excel Modeling: 56.36 Accuracy (%) (#11/20)
• Muse Spark 1.1 on Vals AI Finance Agent v2: 57.21 Accuracy (%) (#2/33)
• Muse Spark 1.1 on Vals AI GPQA: 91.16 Accuracy (%) (#13/121)
• Muse Spark 1.1 on Vals AI Legal Research Bench: 37.98 Accuracy (%) (#7/18)
• Muse Spark 1.1 on Vals AI LegalBench: 84.98 Accuracy (%) (#15/123)
• Muse Spark 1.1 on Vals AI LiveCodeBench: 85.91 Accuracy (%) (#18/127)
• Muse Spark 1.1 on Vals AI MMLU-Pro: 88.73 Accuracy (%) (#10/119)
• Muse Spark 1.1 on Vals AI MMMU: 86.59 Accuracy (%) (#11/81)
• Muse Spark 1.1 on Vals AI MortgageTax: 66.14 Accuracy (%) (#30/84)
• Muse Spark 1.1 on Vals AI Multimodal Index: 66.74 Accuracy (%) (#7/25)
• Muse Spark 1.1 on Vals AI ProgramBench: 16.16 Raw Pass Rate (%) (#24/28)
• Muse Spark 1.1 on Vals AI ProofBench: 39.0 Accuracy (%) (#13/49)
• Muse Spark 1.1 on Vals AI Public Benefits Bench: 61.3 Accuracy (%) (#9/16)
• Muse Spark 1.1 on Vals AI SAGE: 45.58 Accuracy (%) (#22/66)
• Muse Spark 1.1 on Vals AI SWE-bench Verified: 82.0 Resolved (%) (#8/69)
• Muse Spark 1.1 on Vals AI Terminal-Bench 2.1: 69.29 Accuracy (%) (#11/41)
• Muse Spark 1.1 on Vals AI Vibe Code Bench: 72.16 Accuracy (%) (#7/72)

New #1 Leaders (36)

• VoxelBench (Rating): GPT-5.6 (2313.0) beat Claude Fable 5 (Max) (2183.0) by 130.0.
• SEAL - VisualToolBench (VTB) (Score): Muse Spark 1.1 (44.77) beat GPT-5.4 (29.17) by 15.6.
• Vals AI IOI (Accuracy (%)): GPT-5.6 (86.67) beat Claude Fable 5 (72.25) by 14.42.
• SEAL - SWE Atlas - Codebase QnA (Score): Claude Opus 4.8 (Claude Code) xhigh (57.26) beat Claude Opus 4.8 (Claude Code) (48.79) by 8.47.
• Vals AI CyberBench (Accuracy (%)): GPT-5.6 (88.14) beat GPT-5.5 (80.51) by 7.63.
• Vals AI Harvey Legal Agent Bench (Accuracy (%)): muse-spark-1.1 (20.0) beat Grok 4.5 (12.92) by 7.08.
• Vals AI Terminal-Bench 2.1 (Accuracy (%)): GPT-5.6 (85.77) beat Claude Fable 5 (80.52) by 5.25.
• SEAL - Professional Reasoning Benchmark - Legal (Score): Muse Spark 1.1 (57.05) beat Muse Spark (52.29) by 4.76.
• SEAL - MCP Atlas (Score): Muse Spark 1.1 (88.1) beat Gemini 3.5 Flash (High) (83.6) by 4.5.
• SEAL - SWE-Bench Pro (Private Dataset) (Score): Muse Spark 1.1 (51.5) beat Claude Opus 4.6 (Thinking) (47.1) by 4.4.
• Vals AI Legal Research Bench (Accuracy (%)): GPT-5.6 (48.08) beat Claude Opus 4.8 (43.75) by 4.33.
• Gert Labs Rankings (GScore (%)): GPT-5.6 (73.77) beat Claude Fable 5 (69.58) by 4.19.
• KernelBench Hub - Hard (Best % of Hardware Roofline): GPT-5.6 (30.0) beat GLM-5.2 (26.0) by 4.0.
• AA Omniscience - Software Engineering (SWE) - Kotlin (Accuracy (%)): GPT-5.6 (xHigh) (94.0) beat GPT-5.3 Codex (xHigh) (90.0) by 4.0.
• AA Omniscience - Software Engineering (SWE) - PHP (Accuracy (%)): GPT-5.6 (96.0) beat GPT-5.5 (Low) (92.0) by 4.0.
• AA CritPt (Accuracy (%)): GPT-5.6 (32.29) beat Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) (28.57) by 3.72.
• AA Terminal-Bench Hard (Accuracy (%)): GPT-5.6 (65.91) beat Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) (62.88) by 3.03.
• Vals AI Excel Modeling (Accuracy (%)): GPT-5.6 (72.34) beat Claude Opus 4.8 (69.37) by 2.97.
• DeepSWE (Pass@1 (%)): GPT-5 (72.7) beat Claude Fable 5 (69.9) by 2.8.
• SEAL - SWE-Bench Pro (Public Dataset) (Score): Muse Spark 1.1 (61.5) beat GPT-5.4 (xHigh) (59.1) by 2.4.
• PM-LLM-Benchmark (Score): GPT-5.6 (xHigh) (40.2) beat Kimi K2.6 (37.8) by 2.4.
• Vals AI TaxEval v2 (Accuracy (%)): muse-spark-1.1 (79.72) beat Muse Spark (77.68) by 2.04.
• AA Omniscience - Software Engineering (SWE) - HTML (Accuracy (%)): GPT-5.6 (92.0) beat GPT-5.5 (Medium) (90.0) by 2.0.
• AA Omniscience - Health (Accuracy (%)): GPT-5.6 (High) (50.7) beat Grok Build 0.1 (48.9) by 1.8.
• SEAL - Professional Reasoning Benchmark - Finance (Score): Muse Spark 1.1 (55.01) beat Claude Opus 4.6 (Non-reasoning) (53.28) by 1.73.
• FrontierMath - Tiers 1-3 (v2) (Accuracy (%, 285 private v2 problems)): GPT-5.6 (89.12) beat GPT-5.5 Pro (xHigh) (87.72) by 1.4.
• Vals AI SWE-bench Verified (Resolved (%)): GPT-5.6 (96.2) beat Claude Fable 5 (95.0) by 1.2.
• AA Omniscience - Software Engineering (SWE) - TypeScript (Accuracy (%)): GPT-5.6 (92.22) beat GPT-5.5 (xHigh) (91.11) by 1.11.
• WorldScore - Static (Static Score): World Dreamer (84.52) beat EvoPhys-World (83.45) by 1.07.
• Vals AI ProgramBench (Raw Pass Rate (%)): GPT-5.6 (77.64) beat Claude Fable 5 (76.8) by 0.84.
• WorldScore (Average Static/Dynamic Score): World Dreamer (79.44) beat EvoPhys-World (78.62) by 0.82.
• AutomationBench (Pass Rate (%)): GPT-5.6 (18.1) beat Claude Fable 5.0 (Max) (17.4) by 0.7.
• Tau3 Banking (Success Rate (%)): GPT-5.6 (32.99) beat Grok 4.5 (High) (32.58) by 0.41.
• SEAL - MultiNRC (Score): Muse Spark 1.1 (65.59) beat GPT-5 Pro (65.2) by 0.39.
• Vals AI MedScribe (Accuracy (%)): muse-spark-1.1 (88.89) beat Claude Fable 5 (88.52) by 0.37.
• WorldScore - Dynamic (Dynamic Score): World Dreamer (74.35) beat Inspatio-World (74.1) by 0.25.