Claude Sonnet 5.5 vs GLM 4.7 (Venice)
On the 15 use cases both models ran, GLM 4.7 (Venice) won 12 to 3 (better outcome first, then lower cost).
Pass rate: Claude Sonnet 5.5 100% (15 runs) vs GLM 4.7 (Venice) 80% (15 runs). Average cost per run: $0.06 vs $0.03.
| Metric | Claude Sonnet 5.5 | GLM 4.7 (Venice) |
|---|---|---|
| Runs on shared use cases | 15 | 15 |
| Pass rate | 100% (15/15) | 80% (12/15) |
| Avg cost per run | $0.06 | $0.03 |
| Avg wall time | 18s | 6m |
| Step-ins per run | 0.27 | 0.27 |
| Use cases won | 3 | 12 |
Use case by use case
- Backtest a moving-average crossover on a bundled synthetic price file: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Boat fuel range with a reserve over a planned trip: Claude Sonnet 5.5 better (pass 100% vs 0%)
- Catch log checked against a made-up fishing regulations table: Claude Sonnet 5.5 better (pass 100% vs 0%)
- Compute meeting-room utilisation from a versioned booking export across time zones and clock changes: Claude Sonnet 5.5 better (pass 100% vs 0%)
- Dead reckoning, estimated position with current, and computed set and drift: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Debt avalanche vs snowball schedule to the cent: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Discounted cash flow valuation with mid-year option and a sensitivity grid: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Progressive tax on fictional brackets, and pre-tax vs after-tax savings comparison: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Read a synthetic annual-report excerpt and compute financial ratios: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Render per-environment deploy configs when the payment credential is missing: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Retirement drawdown with a 4% rule, inflation raises and a skip-raise-after-loss guardrail: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Schedule a cross-time-zone meeting in a DST-mismatch week and send invites only after sign-off: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Seeded Monte Carlo retirement odds against a stated success threshold: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Split a campaign budget across channels whose response only kicks in above a minimum spend: GLM 4.7 (Venice) better (pass 100% vs 100%)
- Work backward from a retirement budget to a savings target, contributions and earliest retirement age: GLM 4.7 (Venice) better (pass 100% vs 100%)
Only use cases both models ran are compared; all settings of each model are grouped. How we score.