DECISION & CLASSIFICATION / Bespoke Labs
Bespoke Nimble
A decision model and public training recipe for teams exploring their own data, evaluations and deployment.
ACCESS & CAPABILITIES
How it fits your workflow
Apple Silicon or NVIDIA GPU. The project publishes data curation, training and evaluation material; compatibility beyond the documented operations is not assumed.
- Documented judgment types
- Choose an option · Yes / no probability
- Languages
- Cross-language accuracy is not verified in this catalog.
- Context and input limits
- The September 24 update advertises 8,192-token context and up to 255 choices; keep the exact checkpoint with results.
- Jev-compatible interface
- Not verified
PRICE & PROBABILITIES
Hosted price unverified
No hosted rate verified for this entry. Self-hosting needs compute, capacity and maintenance; downloadable weights do not make inference free.
Interpret the output
The project has revised its temperature settings. Re-evaluate thresholds when changing checkpoint or temperature.
Check before choosing
Early research implementation; deployment and evaluation details have changed between releases.
This entry records choice and boolean tasks explicitly documented in the repository, without assuming feature parity with Jev.
No shared benchmark measured yet
Our accuracy, calibration, p50/p95 latency and cost per thousand decisions are not measured. This is a source-reviewed capability profile. Compare the same task, model version, language, input size and serving setup before ranking results.
Read the selection and evaluation guide →Official sources & version notes
Bespoke Labs · Nimble repository ↗ Checked 2026-10-05