Introducing Parallel Decisions API: the first decision engine with world knowledge built in. Learn more.[Learn more.]
Search Capability Leaderboard

Beam

Reflection AIOpen weights

Evaluated

#49 in Search Intelligence

Search Intelligence Score

56.0

+27.2 lift from search

050100
Lift from search
+27.2
28.8 without search
Cost per 1K tasks
$161
Below the median score of 66.0, so not ranked on cost

Accuracy by benchmark

Accuracy on each eval, with search (solid) and without (light).

  • DSQA68.9 vs 38.4
  • HLE39.0 vs 38.0
  • WISER60.0 vs 10.0

Cost per 1K tasks

How the cost splits between model inference, search calls, and page extraction.

Inference
$133
83%
Search
$8.38
5%
Extract
$19.3
12%

Compared with

Other models from the same provider alongside models that score closest.

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.

# Beam / Search Capability Leaderboard

  • Search Intelligence Score: 56.0 (#49 in Search Intelligence)
  • Without search: 28.8 / Lift from search +27.2
  • Cost per 1K tasks: $161 (below the median score of 66.0, so not ranked on cost)
  • Evaluated:
  • Lab: Reflection AI / Weights: Open weights / Model id: reflection/beam

## Accuracy by benchmark

SuiteWith searchWithout search
DSQA68.938.4
HLE39.038.0
WISER60.010.0

## Compared with

ModelScoreLiftCost per 1K tasks
Beam56.0+27.2$161
GLM-5.254.3+35.9$66.6
Hunyuan 358.2+35.6$28.4
DeepSeek V4 Pro58.4+27.9$110
Gemini 3 Flash58.6+24.5$114

Leaderboard last updated . Methodology: /leaderboard#methodology