Shareable benchmark table
EVMBench Detect recall by agent system, four arms (never pooled)
Reference-detection recall on the same 12 frozen audits. Ox Alpha (OpenCode) and GLM-5.3 (ZCode) ran 1 run/audit scored by themselves; GPT-5.6 Sol and Grok 4.6 Build ran 3 runs/audit with two independent reviewers plus adjudication.
Table 22 · Adjacent cross-system
EVMBench Detect recall by agent system, four arms (never pooled)
| Agent system | Runs | Refs detected | Micro recall | Macro recall | Award-weighted | Precision (closed) | F1 (closed) | Median wall |
|---|---|---|---|---|---|---|---|---|
| Ox Alpha · OpenCode 1.18.20 (self-scored) | 12 | 4/43 | 9.3% | 7.6% | 1.03% | 25.0% | 13.6% | 2m 50s |
| GPT-5.6 Sol · High (Codex CLI) | 36 | 52/129 | 40.3% | 46.1% | 6.81% | 50.5% | 44.8% | 4m 24s |
| Grok 4.6 Build · High (Grok CLI) | 36 | 47/129 | 36.4% | 41.8% | 4.78% | 45.6% | 40.5% | 10m 25s |
| GLM-5.3 · ZCode CLI 3.8.1 (self-scored) | 12 | 5/43 | 11.6% | 12.1% | 1.91% | 29.4% | 16.7% | 1m 09s |
Adjacent descriptive results under different protocols: replicate counts (1 vs 3), scorer independence, harness CLI, and session structure all differ. Ox and GLM columns are single-reviewer self-scored reconstructions, not official EVMbench grading. The Ox figure is the clean-room v3 run (pre-freeze blindness held); two earlier answer-key-contaminated attempts were discarded and archived. Never pooled into one leaderboard.
