Skip to content

Kev 27B

Jared PalmerDecision modelOpen weights

Kev 27B by Jared Palmer is a decision model: it answers typed questions with a probability for each allowed answer instead of generating text. It scores 58.8 on the Jev Decision Index (rank 13).

Pricing

No hosted price is listed. The weights are open: the cost is your own compute.

Specifications

Provider
Jared Palmer
Parameters
28B
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/Qwen3.8-27B
Jev-compatible request
Yes
Runs on CPU
No
License
apache-2.0
Hugging Face
jaredpalmer/kev-27b

Benchmarks

Jev Decision IndexECE 0.022
58.8

Jev Decision Index: public 56.7, ECE 0.022, Brier 0.359. Lower ECE means its probabilities can be taken at face value.

Self-reported by the publisher

The model's own claims, often on its own sets or a self-run copy of a public suite. Not independently verified and never ranked alongside measured results.

breadth-v1 test: 10 of 14 held-out public datasets (2,489 quSelf-reported
0.832
hard-v1 + devtools-v1 + documents-v1 test (2,795 questions)Self-reported
0.889
tasksource-heldout-v1 test: 17 of 24 held-out task familiesSelf-reported
0.795
transfer-v4 test: six never-trained public sources and held-Self-reported
0.889

Speed & latency

Median latency per call (Decision Index)
135 ms

Compare Kev 27B with…

FAQ

How much does Kev 27B cost?

Kev 27B has no hosted price listed; its weights are open, so the cost is the compute you run it on.

What questions can Kev 27B answer?

Kev 27B answers choice (one of n options), score (ordered rubric), yes / no questions, returning a probability for each allowed answer.

Is Kev 27B well calibrated?

On the Jev Decision Index its expected calibration error (ECE) is 0.022; below about 0.05, a stated 80% confidence is right roughly 80% of the time.

Is Kev 27B open weight, and can it run on a CPU?

Yes, Kev 27B 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, Benchmark Heaven JevBench, Hugging Face). Pricing is per 1M tokens (USD); confirm with the provider before production use.