Show HN: Snap – local-first, open-weight 2B decision engine by emnlmn

Kev is a family of small Jev-architecture decision models (0.8B, 4B, 9B) built on Qwen3.5 and released under Apache-2.0 with training code and frozen eval suites. The 9B model runs in bf16 on a 32GB Mac, and the models are API-compatible with TypeSafe's System One, so its Python SDK works against a local Kev server unchanged. The release targets LLM-as-judge eval pipelines, and the web playground demonstrates option-order bias that judge models commonly exhibit.

Releases·Model Release

Trained locally: ultra-fast 0.8B/2B System 1 decision models that match Jev on benchmarks and Doom, ~30 ms per decision (open weights)

A developer fine-tuned Qwen3.5 0.8B/2B and Gemma models into zero-shot classifiers ('Jeff models') that return calibrated probabilities in one forward pass, releasing Apache 2.0 weights on HuggingFace. The 2B scores 83.1% on a five-benchmark panel versus Jev's published 83.0%, while deciding in ~28 ms locally. Training used only local hardware and synthetic data from a local teacher model, replicating TypeSafe's Jev experiment with small models.

A developer fine-tuned Qwen3.5 0.8B/2B and Gemma models into zero-shot classifiers ('Jeff models') that return calibrated probabilities in one forward pass, releasing Apache 2.0 weights on HuggingFace. The 2B scores 83.1% on a five-benchmark panel versus Jev's published 83.0%, while deciding in ~28 ms locally. Training used only local hardware and synthetic data from a local teacher model, replicating TypeSafe's Jev experiment with small models.

Researchers released this-that-model-1.0, a 2B-parameter typed decision model that reads answers directly from hidden states, outputs one of n declared options plus a thresholdable number, and generates zero tokens. It decides in 30.9 ms on one laptop GPU versus 8758 ms for a frontier API call, and scores 0.941 (Brier 0.042) on a third party's 68-question cohort versus 0.765 (Brier 0.133) for hosted service Jev. It is open-sourced on Hugging Face, targeting software branches that need decisions, not prose.

Products·launched

Jevless: Jev-style typed decisions (Choice/Noul/Score) from any logprobs-exposing model API

Jevless is an open-source tool that produces Jev-style typed decisions — Choice, Noul, and Score outputs — using any LLM whose API exposes logprobs, rather than requiring a single supported backend. It also ships a local /v1/systemone-compatible server. The author built it to keep Jev's low cost, speed, and bias-canceling accuracy while working across as many model backends as possible.

Jevless is an open-source tool that produces Jev-style typed decisions — Choice, Noul, and Score outputs — using any LLM whose API exposes logprobs, rather than requiring a single supported backend. It also ships a local /v1/systemone-compatible server. The author built it to keep Jev's low cost, speed, and bias-canceling accuracy while working across as many model backends as possible.