Search Capability Leaderboard
Claude Opus 5.5
Anthropic
1Gold · Search Intelligence
#12 in Search Efficiency
Search Intelligence Score
75.5
+32.9 lift from search
050100
- Lift from search
- +32.9
- 42.7 without search
- Cost per 1K tasks
- $1,656
- #12 in Search Efficiency
- Time per task
- 525s
- Search plus inference time to complete one task
Accuracy by benchmark
Accuracy on each eval, with search (solid) and without (light).
- DSQA88.6 vs 57.0
- HLE57.0 vs 47.0
- WISER81.0 vs 24.0
Cost per 1K tasks
How the cost splits between model inference, search calls, and page extraction.
- Inference
- $1,643
- 99%
- Search
- $3.39
- 0%
- Extract
- $10.6
- 1%
Compared with
Other models from the same provider alongside models that score closest.
- Opus 5.575.5Cost $1,656 · Lift +32.9 · 525s
- Opus 571.2Cost $1,437 · Lift +30.8 · 492s
- Fable 5.169.3Cost $3,101 · Lift +24.7 · 536s
- Sonnet 560.4Cost $1,000 · Lift +38.1 · 775s
- GPT-6 Astra70.8Cost $401 · Lift +27.9 · 83.7s
| Model | Score | Lift | Cost per 1K tasks | Time per task |
|---|---|---|---|---|
| Claude Opus 5.5 | 75.5 | +32.9 | $1,656 | 525s |
| Opus 5Claude Opus 5 | 71.2 | +30.8 | $1,437 | 492s |
| Fable 5.1Claude Fable 5.1 | 69.3 | +24.7 | $3,101 | 536s |
| Sonnet 5Claude Sonnet 5 | 60.4 | +38.1 | $1,000 | 775s |
| GPT-6 AstraGPT-6 Astra | 70.8 | +27.9 | $401 | 83.7s |
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 22, 2026. Full methodology.
# Claude Opus 5.5 · Search Capability Leaderboard
- Search Intelligence Score: 75.5 (#1 in Search Intelligence)
- Without search: 42.7 · Lift from search +32.9
- Cost per 1K tasks: $1,656 (#12 in Search Efficiency)
- Time per task: 525s
- Medals: Gold, Search Intelligence
- Lab: Anthropic · Model id: claude-opus-5.5
## Accuracy by benchmark
| Suite | With search | Without search |
|---|---|---|
| DSQA | 88.6 | 57.0 |
| HLE | 57.0 | 47.0 |
| WISER | 81.0 | 24.0 |
## Compared with
| Model | Score | Lift | Cost per 1K tasks | Time per task |
|---|---|---|---|---|
| Claude Opus 5.5 | 75.5 | +32.9 | $1,656 | 525s |
| Claude Opus 5 | 71.2 | +30.8 | $1,437 | 492s |
| Claude Fable 5.1 | 69.3 | +24.7 | $3,101 | 536s |
| Claude Sonnet 5 | 60.4 | +38.1 | $1,000 | 775s |
| GPT-6 Astra | 70.8 | +27.9 | $401 | 83.7s |
Latest update September 22, 2026. Methodology: /leaderboard#methodology