Needle 3, options as tools (post-hoc adapter mode)
Cactus Compute · Apache-2.0 (model and package) · open (code and weights) · small-tool-model
JevBench v1.3.0 score
1.1
Not ranked. Listed as a partial.
73.3 points behind Jev 1.13.0's 74.4.
Axes
- Intelligence
- 13.5
- Calibration
- —
- Speed
- 52.8
- Cost
- 65.3
Cost
$0.014 per 1,000 decisions (estimated).
github.com/cactus-compute/needle
See every JevBench v1.3.0 system, method and scoring on the full leaderboard.