gpt-5.6-luna
gpt-5.6-luna has 5 tested configurations. Its highest published overall score is 52.35 at xhigh. Inspect dimension scores, evaluation time, and reference cost below.
Overall score52.35
gpt-5.6-luna · xhigh
- Backend engineering
- 56
- Frontend interaction
- 83
- Knowledge & Reasoning
- 25
- Visual aesthetics
- 66
- Maintainability
- 59
- Evaluation time
- 75m 23s
- Reference cost
- $0.16
Complete evaluation · API reference cost
Results by reasoning effort
Swipe to see all results →
| Model / effort | Overall | Backend engineering | Frontend interaction | Knowledge & Reasoning | Visual aesthetics | Maintainability | Total eval. time | Total ref. cost | Compare |
|---|---|---|---|---|---|---|---|---|---|
| — | 32 | 18 | 5 | — | — | 52m 31s | $0.067 | ||
| — | 51 | 13 | 25 | — | — | 18m 22s | $0.084 | ||
| 38.30 | 49 | 52 | 15 | 55 | 46 | 32m 18s | $0.14 | ||
| 52.35 | 56 | 83 | 25 | 66 | 59 | 75m 23s | $0.16 | ||
| 52.30 | 65 | 78 | 25 | 59 | 61 | 86m 14s | $0.35 |
Score breakdown
- Overall score
- 52.35
- Evaluation time
- 75m 23s
- Reference cost
- $0.16
Service logic, error handling, and test verification.
Contracts and boundaries8 / 20
API contracts, input validation, and boundary behavior.
Retries and failures15 / 20
Error recovery, retry conditions, and failure handling.
CI and adversarial review13 / 20
Verification plans, risk detection, and audit evidence.
Transactions and consistency12 / 20
State transitions, concurrency control, and data consistency.
Caching and regression8 / 20
Cache invalidation, data updates, and regression coverage.
What changes with higher reasoning effort
Compare overall scores, cost, and time across neighboring tested efforts.
high xhigh
- Overall
- +14.1 pts 38.30 → 52.35
- Reference cost
- +$0.027 $0.14 → $0.16
- Evaluation time
- +133.4% 32m 18s → 75m 23s
Data snapshot published · 10/27/2026Scoring & cost notes