Hyperspace AI releases jevcache, a Rust decision cache for Jev-class models that memoizes (model, schema, state) decisions locally

Releases·Tool Release

LLM2Jev turns local LLMs into Jev-style structured decision engines with prefill-only binary inference

A new open-source project, LLM2Jev, adapts local Hugging Face-compatible causal language models into Jev-style structured decision engines, emitting typed Choice, Score, and Noul answers with probabilities. It uses prefill-only binary inference — scoring candidates with yes/no logit judgments without generating answer tokens — and leverages SGLang's Radix Cache for shared-prefix efficiency. The project is independent and unaffiliated with Jev or TypeSafe.

A new open-source project, LLM2Jev, adapts local Hugging Face-compatible causal language models into Jev-style structured decision engines, emitting typed Choice, Score, and Noul answers with probabilities. It uses prefill-only binary inference — scoring candidates with yes/no logit judgments without generating answer tokens — and leverages SGLang's Radix Cache for shared-prefix efficiency. The project is independent and unaffiliated with Jev or TypeSafe.

Products·launched

kevMac: native macOS app for running Jared Palmer's kev (Jev) decision model locally, offline

kevMac is a lightweight native macOS app that wraps Jared Palmer's open-source kev model, a Jev-compatible decision model built on Qwen. It lets users run Jev-style System One decision models completely locally and offline without writing JSON or Python, returning structured decisions instead of generating text token-by-token.

kevMac is a lightweight native macOS app that wraps Jared Palmer's open-source kev model, a Jev-compatible decision model built on Qwen. It lets users run Jev-style System One decision models completely locally and offline without writing JSON or Python, returning structured decisions instead of generating text token-by-token.

Social·Community

awesome-jev: curated list of projects built on TypeSafe AI's Jev typed-decision model

A curated GitHub repository, awesome-jev, aggregates public projects, integrations, and discussions built on Jev, TypeSafe AI's System One model for typed decisions. Jev is not a chat model: it takes unstructured state plus a typed question and returns a typed decision — a choice, score, or boolean with confidence — positioning it as a drop-in decision layer for classification, routing, scoring, verification, and agent guardrails. The list gives practitioners a fast-scanning field guide to which decision patterns transfer across industries.

A curated GitHub repository, awesome-jev, aggregates public projects, integrations, and discussions built on Jev, TypeSafe AI's System One model for typed decisions. Jev is not a chat model: it takes unstructured state plus a typed question and returns a typed decision — a choice, score, or boolean with confidence — positioning it as a drop-in decision layer for classification, routing, scoring, verification, and agent guardrails. The list gives practitioners a fast-scanning field guide to which decision patterns transfer across industries.

Products·launched

CacheCraft — a bench for testing semantic cache strategies on recorded request traces

CacheCraft is a semantic cache strategy benchmarking tool. It replays recorded traces of real LLM requests through candidate caching policies and reports which policy would have paid off in latency avoided and quality risk accepted. It targets teams deciding whether and how to enable semantic caching for LLM workloads.

CacheCraft is a semantic cache strategy benchmarking tool. It replays recorded traces of real LLM requests through candidate caching policies and reports which policy would have paid off in latency avoided and quality risk accepted. It targets teams deciding whether and how to enable semantic caching for LLM workloads.

Products·launched

INVAR: local LLM runner with verifiable inference receipts

INVAR is a local LLM runner that wraps llama.cpp and generates a cryptographic receipt for every inference, enabling later verification of which model, prompt, and settings produced an output. It solves the problem of proving provenance and reproducibility for AI-generated content, useful for developers and professionals who need audit trails. It is free, open-source (Apache-2.0), and runs entirely locally with no telemetry.

INVAR is a local LLM runner that wraps llama.cpp and generates a cryptographic receipt for every inference, enabling later verification of which model, prompt, and settings produced an output. It solves the problem of proving provenance and reproducibility for AI-generated content, useful for developers and professionals who need audit trails. It is free, open-source (Apache-2.0), and runs entirely locally with no telemetry.