Search Intelligence Score
70.3
+35.3 lift from search
050100
- Lift from search
- +35.3
- 35.0 without search
- Cost per 1K tasks
- $301
- #11 in Search Efficiency
- Time per task
- 34.7s
- Search plus inference time to complete one task
Accuracy by benchmark
Accuracy on each eval, with search (solid) and without (light).
- DSQA74.0 vs 17.0
- HLE64.0 vs 63.0
- WISER73.0 vs 25.0
Cost per 1K tasks
How the cost splits between model inference, search calls, and page extraction.
- Inference
- $295
- 98%
- Search
- $2.66
- 1%
- Extract
- $2.82
- 1%
InferenceSearchExtract
Compared with
Other models from the same provider alongside models that score closest.
| Model | Score | Lift | Cost per 1K tasks | Time per task |
|---|---|---|---|---|
| GPT-6 (Astra) | 70.3 | +35.3 | $301 | 34.7s |
| GPT-5.6 SolGPT-5.6 Sol | 66.7 | +35.3 | $217 | 52.9s |
| GPT-5.6 LunaGPT-5.6 Luna | 54.7 | +32.3 | $23.0 | 59.4s |
| DS Flash 4.1DeepSeek V4.1 Flash | 65.3 | +45.0 | $44.8 | 124s |
| Muse Spark 1.3Muse Spark 1.3 | 65.3 | +34.3 | $98.2 | 66.2s |
Search Intelligence Score averages accuracy on DSQA, HLE, and WISER with equal weight; cost and time are averaged the same way, with cost shown per 1,000 tasks. Latest update September 14, 2026. Full methodology.
# GPT-6 (Astra) · Search Capability Leaderboard
- Search Intelligence Score: 70.3 (#1 in Search Intelligence)
- Without search: 35.0 · Lift from search +35.3
- Cost per 1K tasks: $301 (#11 in Search Efficiency)
- Time per task: 34.7s
- Medals: Gold, Search Intelligence
- Lab: OpenAI · Model id: gpt-6-astra
## Accuracy by benchmark
| Suite | With search | Without search |
|---|---|---|
| DSQA | 74.0 | 17.0 |
| HLE | 64.0 | 63.0 |
| WISER | 73.0 | 25.0 |
## Compared with
| Model | Score | Lift | Cost per 1K tasks | Time per task |
|---|---|---|---|---|
| GPT-6 (Astra) | 70.3 | +35.3 | $301 | 34.7s |
| GPT-5.6 Sol | 66.7 | +35.3 | $217 | 52.9s |
| GPT-5.6 Luna | 54.7 | +32.3 | $23.0 | 59.4s |
| DeepSeek V4.1 Flash | 65.3 | +45.0 | $44.8 | 124s |
| Muse Spark 1.3 | 65.3 | +34.3 | $98.2 | 66.2s |
Latest update September 14, 2026. Methodology: /leaderboard#methodology