Search Intelligence Score
49.0
+34.7 lift from search
050100
- Lift from search
- +34.7
- 14.3 without search
- Cost per 1K tasks
- $41.0
- Below the median score of 51.2, so not ranked on cost
- Time per task
- 239s
- Search plus inference time to complete one task
Accuracy by benchmark
Accuracy on each eval, with search (solid) and without (light).
- DSQA51.0 vs 13.0
- HLE42.0 vs 21.0
- WISER54.0 vs 9.0
Cost per 1K tasks
How the cost splits between model inference, search calls, and page extraction.
- Inference
- $26.8
- 65%
- Search
- $9.29
- 23%
- Extract
- $4.85
- 12%
InferenceSearchExtract
Compared with
Other models from the same provider alongside models that score closest.
| Model | Score | Lift | Cost per 1K tasks | Time per task |
|---|---|---|---|---|
| Hunyuan 3 | 49.0 | +34.7 | $41.0 | 239s |
| MiniMax M3MiniMax M3 | 49.0 | +29.0 | $39.9 | 156s |
| DS FlashDeepSeek V4 Flash | 50.0 | +35.0 | $18.8 | 150s |
| DS ProDeepSeek V4 Pro | 50.0 | +28.0 | $102 | 205s |
| DS Pro 0813DeepSeek V4 Pro (0813) | 50.3 | +27.0 | $108 | 213s |
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.
# Hunyuan 3 · Search Capability Leaderboard
- Search Intelligence Score: 49.0 (#18 in Search Intelligence)
- Without search: 14.3 · Lift from search +34.7
- Cost per 1K tasks: $41.0 (below the median score of 51.2, so not ranked on cost)
- Time per task: 239s
- Lab: Tencent · Model id: tencent/hy3
## Accuracy by benchmark
| Suite | With search | Without search |
|---|---|---|
| DSQA | 51.0 | 13.0 |
| HLE | 42.0 | 21.0 |
| WISER | 54.0 | 9.0 |
## Compared with
| Model | Score | Lift | Cost per 1K tasks | Time per task |
|---|---|---|---|---|
| Hunyuan 3 | 49.0 | +34.7 | $41.0 | 239s |
| MiniMax M3 | 49.0 | +29.0 | $39.9 | 156s |
| DeepSeek V4 Flash | 50.0 | +35.0 | $18.8 | 150s |
| DeepSeek V4 Pro | 50.0 | +28.0 | $102 | 205s |
| DeepSeek V4 Pro (0813) | 50.3 | +27.0 | $108 | 213s |
Latest update September 14, 2026. Methodology: /leaderboard#methodology