GPT-6 Luna Decisions
OpenAI · openai/gpt-6-luna-decisions · added 20 Oct 2026
GPT-6 Luna Decisions by OpenAI ranks #8 of 12 AI decision models overall on DecisionBench: #11 of 12 on Sudoku, #9 of 12 on Triage, #3 of 12 on Robustness. View on OpenRouter.
- Overall
- #8 of 12
- Average rank
- 7.7
- Sudoku
- #11 of 12
- Triage
- #9 of 12
- Robustness
- #3 of 12
Sudoku Logical reasoning
Leaderboard →- Rank
- #11 of 12
- Accuracy
- 74.0%
- Solved
- 19/20
- Mistakes / puzzle
- 17.1
- Solve time / puzzle
- 14s
- Cost / puzzle
- $0.0188
By difficulty
| Difficulty | Rank | Accuracy | Solved | Mistakes | Solve time | Cost |
|---|---|---|---|---|---|---|
| Easy | #10 | 85.1% | 5/5 | 7.2 | 8.3s | $0.009636 |
| Medium | #12 | 74.7% | 5/5 | 16.6 | 13s | $0.0171 |
| Hard | #12 | 73.8% | 4/5 | 17.0 | 13s | $0.0190 |
| Expert | #11 | 67.3% | 5/5 | 27.8 | 21s | $0.0293 |
Every race (20)
| Puzzle | Mistakes | Calls | Solve time | Cost | Status |
|---|---|---|---|---|---|
| easy-01 | 5 | 19 | 8.8s | $0.008555 | Solved |
| easy-02 | 10 | 22 | 9.4s | $0.0123 | Solved |
| easy-03 | 5 | 17 | 7.0s | $0.008096 | Solved |
| easy-04 | 9 | 21 | 8.3s | $0.0105 | Solved |
| easy-05 | 7 | 18 | 8.2s | $0.008745 | Solved |
| medium-01 | 21 | 31 | 13s | $0.0193 | Solved |
| medium-02 | 10 | 24 | 9.8s | $0.0138 | Solved |
| medium-03 | 22 | 33 | 13s | $0.0206 | Solved |
| medium-04 | 19 | 30 | 15s | $0.0183 | Solved |
| medium-05 | 11 | 24 | 12s | $0.0137 | Solved |
| hard-01 | 19 | 23 | 11s | $0.0176 | Error |
| hard-02 | 15 | 31 | 12s | $0.0192 | Solved |
| hard-03 | 17 | 30 | 14s | $0.0187 | Solved |
| hard-04 | 22 | 37 | 15s | $0.0239 | Solved |
| hard-05 | 12 | 27 | 11s | $0.0159 | Solved |
| expert-01 | 27 | 43 | 20s | $0.0290 | Solved |
| expert-02 | 26 | 41 | 20s | $0.0281 | Solved |
| expert-03 | 28 | 44 | 21s | $0.0296 | Solved |
| expert-04 | 31 | 46 | 23s | $0.0308 | Solved |
| expert-05 | 27 | 43 | 23s | $0.0288 | Solved |
Triage Following written rules
Leaderboard →- Rank
- #9 of 12
- Accuracy
- 85.0%
- Team
- 100.0%
- Refund
- 98.3%
- Urgency
- 56.7%
- Confidence gap
- 0.074
60 tickets · 3.7% of mistakes made while at least 90% sure · $0.0916 per 1,000 tickets
Robustness Consistency under pressure
Leaderboard →- Rank
- #3 of 12
- Score
- 99.7%
- Same if re-asked
- 100.0%
- Changed by reorder
- 0.0%
- In a list of 40
- 87% / 80%
- Obeyed planted text
- 1.7%
20 tickets × 6 checks · changed by rename 0.0% · accuracy with planted text 73.3%