Decido

Decido is an AI model and API server that evaluates documents and answers questions on a laptop CPU.

Decido provides calibrated yes/no, pick-one, and rating decisions about documents using a fine-tuned Gemma model running on llama.cpp. It takes text, JSON, or lists alongside typed questions, returning probabilities for every answer in a Jev-compatible wire format. The system operates via an HTTP API server that matches TypeSafe Jev endpoints, allowing existing SDKs to switch with a single setting. It is built for developers and researchers who need local, deterministic decision processing without relying on external cloud APIs. The open-source software is licensed under Apache-2.0 and supports macOS and Linux environments with sufficient RAM.

Decido at a glance

Best for: Software engineer, Researcher
Runs on: macOS, Linux, API, self-hosted, open source
Pricing: Free
Released: October 2026
Category: AI research tool
Website: github.com/decido-ai/decido

Key features

Evaluates yes/no, multiple choice, and score questions
Runs locally on standard laptop CPUs using llama.cpp
Provides a Jev-compatible HTTP API server
Supports gzip and zstd request bodies
Accepts text, JSON, or list states up to sixteen thousand tokens
Offers an optional API key authorization bearer
Includes training scripts and evaluation harnesses

Use it for

Answering typed questions about large documents locally
Replacing cloud decision APIs with a self-hosted CPU server
Running automated decision benchmarks and parity tests
Processing structured text and JSON data for probabilities

Alternatives to Decido

SemIf OpenJev — Extracts direct token probabilities without generating text or running JSON repair loops.
Laya — It runs typed decisions in a single forward pass over 100+ languages.
Jiwo — It provides low-latency decision-making with a 50ms median request time.
Vegaml — Runs locally on your hardware with calibrated probabilities and conformal answer sets.
Benchpress — Evaluates document artificial intelligence models against specific user pages for accuracy, cost, and speed.
NanoJev — It runs parallel decisions and outputs probability distributions without output-token decoding.
Kotaemon — It offers an open-source, customizable interface for local and API-based document retrieval.
Console — Combines fast AI decision routing with precise document reading in one API key.
MiniCPM — Offers 2B-class open-source SOTA performance with compact on-device deployment.
Homi — It answers complex audio questions natively with high accuracy.
ODS — Automates complex local AI infrastructure setups with one command.
HyperResearch — Generates verified reports with citations checked against full source texts and a persistent vault.

All Decido alternatives