JPT-4B
JPT-4B by kirp is a decision model: it answers typed questions with a probability for each allowed answer instead of generating text. It scores 41.6 on the Jev Decision Index (rank 76).
Pricing
No hosted price is listed. The weights are open: the cost is your own compute.
Specifications
- Provider
- kirp
- Parameters
- 5B
- Released
- Sep 2026
- Open weights
- Yes
- Question types
- Choice (one of N options), Score (ordered rubric), Yes / no
- Mechanism
- Decoder LM with a decision readout
- Backbone
- Qwen/Qwen3.5-4B-Base
- Jev-compatible request
- Yes
- Runs on CPU
- No
- License
- cc-by-nc-4.0
- Hugging Face
- kirp/jpt-4b
Benchmarks
- Jev Decision IndexECE 0.079
- 41.6
- Jev Decision Index — VisionECE 0.057
- 59.0
Jev Decision Index: public 42.8, ECE 0.079, Brier 0.448. Lower ECE means its probabilities can be taken at face value.
Speed & latency
Compare JPT-4B with…
FAQ
How much does JPT-4B cost?
JPT-4B has no hosted price listed; its weights are open, so the cost is the compute you run it on.
What questions can JPT-4B answer?
JPT-4B answers choice (one of n options), score (ordered rubric), yes / no questions, returning a probability for each allowed answer.
Is JPT-4B well calibrated?
On the Jev Decision Index its expected calibration error (ECE) is 0.079; below about 0.05, a stated 80% confidence is right roughly 80% of the time.
Is JPT-4B open weight, and can it run on a CPU?
Yes, JPT-4B has open weights. It needs a GPU to run at useful speed.
Data as of 2026-10-14. Source: Metatext decision-model catalogue, 2026-10-14 (Jev Decision Index, Enterprise Decision Arena, Hugging Face). Pricing is per 1M tokens (USD); confirm with the provider before production use.