GPT-6 Pro Luna: benchmark results
OpenAI GPT-6 Luna run in Pro reasoning mode (reasoning.mode "pro" on gpt-6-luna; OpenAI offers no separate Pro model slug). Kept separate from standard GPT-6 Luna in benchmark results; GPT-6 Luna was released in the API on September 22, 2026. Provider: OpenAI. Released 2026-09-22. Access: API.
Aggregate Rating 1813 ± 9, rank #72 of 10916 rated models, from 11 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ObviousBench | 100 | Answer pass³ (%) | 96.8 |
| OpenRouter GPQA Diamond | 88.4 | Accuracy (%) | 74.3 |
| Tinybird AI SQL Benchmark - Success Rate | 100 | Questions answered with a valid query within 3 retries (%) | 72 |
| MineBench | 1752 | Elo Rating | 68.4 |
| Bullshit Benchmark | 40 | BS Detection Rate (%) | 66.7 |
| AIMultiple - Text-to-SQL (Adjudicated) | 72.1 | Execution match with jury-credited equivalents and broken go | 41.2 |
| AIMultiple - Text-to-SQL (Strict Execution Match) | 36.5 | Execution match against BIRD gold (%, over correctly routed | 39.4 |
| AIMultiple - Agentic RAG Routing | 71.74 | Routing accuracy (%, 184 hardest of 759 BIRD questions, 11 d | 35 |
| Tinybird AI SQL Benchmark - Exactness | 45.27 | Result exactness vs human reference queries (0-100) | 34.8 |
| OpenRouter Tau2-Bench Airline | 66.7 | Accuracy (%) | 31.7 |
| Tinybird AI SQL Benchmark - First-Attempt Success Rate | 82 | Questions answered with a valid query on the first attempt ( | 21.2 |
Interactive version: theaggregate.ai/model?slug=gpt-6-pro-luna · How It Works · Data refreshed daily, snapshot 2026-11-08.