Lev
Apache-2.0 self-hosted decision engine with a Jev-compatible API that routes between Laya encoders and a llama.cpp model read by token logprobs.
0 members list Lev, unchanged over the last 12 weeks, data as of 16 November 2026
Members who list Lev in their Stack, by week.
Members who list Lev
Be the first to add Lev
Nobody on Stackness runs Lev yet. Add it to your stack and you're the first member running it.
About Lev
Lev is a self-hosted decision engine that provides an API compatible with Jev and handles routing between Laya encoders and a language model powered by llama.cpp. The tool makes inference decisions by examining token log probabilities from the underlying model. It implements the Laya non-autoregressive System 1 approach, which focuses on efficient decision-making without generating text sequentially. Developers use Lev when they need a locally-hosted inference system that can evaluate multiple pathways and select outputs based on probability analysis rather than autoregressive generation.
Lev was developed by jlt-commons and is built in Clojure. The project is distributed under the Apache 2.0 license, making it available as open source software. It integrates with transformer-based models and supports modern architectures like ModernBERT, alongside BLAS acceleration for optimized computation on various hardware platforms.
Checked 11 November 2026.
Nobody on Stackness runs Lev yet.
Sign up and add itPosts about Lev
- Do you need Jev? Typed decisions from an ordinary LLM with one token and logprobs
You can get Jev-style typed decisions from an ordinary LLM by forcing one output token and reading its logprobs. Where that matches Jev, and where it loses.