Search Intelligence Score
61.9
+33.9 lift from search
050100
- Lift from search
- +33.9
- 28.0 without search
- Cost per 1K tasks
- $33.1
- Below the median score of 64.6, so not ranked on cost
Accuracy by benchmark
Accuracy on each eval, with search (solid) and without (light).
- DSQA78.7 vs 45.0
- HLE42.0 vs 31.0
- WISER65.0 vs 8.0
Cost per 1K tasks
How the cost splits between model inference, search calls, and page extraction.
- Inference
- $14.0
- 42%
- Search
- $7.67
- 23%
- Extract
- $11.4
- 34%
Compared with
Other models from the same provider alongside models that score closest.
- GPT-6 Luna61.9Cost $33.1 / Lift +33.9
- GPT-5.6 Luna60.7Cost $36.2 / Lift +33.9
- GPT-6 Sol66.6Cost $183 / Lift +28.7
- GPT-5.6 Sol67.7Cost $269 / Lift +26.3
- GPT-6.1 Terra69.7Cost — / Lift +41.2
| Model | Score | Lift | Cost per 1K tasks |
|---|---|---|---|
| GPT-6 Luna | 61.9 | +33.9 | $33.1 |
| GPT-5.6 LunaGPT-5.6 Luna | 60.7 | +33.9 | $36.2 |
| GPT-6 SolGPT-6 Sol | 66.6 | +28.7 | $183 |
| GPT-5.6 SolGPT-5.6 Sol | 67.7 | +26.3 | $269 |
| GPT-6.1 TerraGPT-6.1 Terra | 69.7 | +41.2 | — |
Search Intelligence Score averages accuracy on DSQA, HLE, and WISER with equal weight; cost is averaged the same way, with cost shown per 1,000 tasks. Evaluated . Leaderboard last updated . Full methodology.
# GPT-6 Luna / Search Capability Leaderboard
- Search Intelligence Score: 61.9 (#34 in Search Intelligence)
- Without search: 28.0 / Lift from search +33.9
- Cost per 1K tasks: $33.1 (below the median score of 64.6, so not ranked on cost)
- Evaluated:
- Lab: OpenAI / Weights: Closed / Model id: gpt-6-luna
## Accuracy by benchmark
| Suite | With search | Without search |
|---|---|---|
| DSQA | 78.7 | 45.0 |
| HLE | 42.0 | 31.0 |
| WISER | 65.0 | 8.0 |
## Compared with
| Model | Score | Lift | Cost per 1K tasks |
|---|---|---|---|
| GPT-6 Luna | 61.9 | +33.9 | $33.1 |
| GPT-5.6 Luna | 60.7 | +33.9 | $36.2 |
| GPT-6 Sol | 66.6 | +28.7 | $183 |
| GPT-5.6 Sol | 67.7 | +26.3 | $269 |
| GPT-6.1 Terra | 69.7 | +41.2 | — |
Leaderboard last updated . Methodology: /leaderboard#methodology