# Decider (slim)

> Decider is a family of open-weight decision models by Mark Marosi (Mapika), from 0.8B to 35B parameters under Apache-2.0. It answers typed yes or no, choice and score questions with probabilities, and runs locally from the decider-ai Python package.

- Full: https://www.anchorterminal.com/tools/decider.md (~5,700 tokens) · this version ~1,330 tokens · JSON https://www.anchorterminal.com/tools/decider.json · canonical https://www.anchorterminal.com/tools/decider
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · Updated: 2026-10-09

**B · 69.5/100 · rank #172 of 842 · #2 in Decision models · not agent-ready · confidence medium**


Assessment: An Apache-2.0 decision model family with a dated changelog, passing CI, 21 package releases since 22 September 2026 and model cards that list measured regressions. One person maintains it, the local server has no authentication option, states over 32,768 tokens are cut without an error, and no security policy is published.

## Facts

- Kind: Model API · vendor: Mark Marosi (Mapika) · category: Decision models · legal entity: not named · provenance 27/100
- Local only (HTTP): pypi `decider-ai`
- Auth: None · pricing: Free · x402: no · licence: Apache-2.0 (code and weights)
- Probe metrics: not measured yet (probes haven't run)
- Models: decider-0.8b, decider-2b v11 (1.9B), decider-4b v2.1 (4.2B) and decider-35b-a3b v1 (34.7B, 3B active) on Qwen3.5 bases. decider-12b v2 and decider-31b on Gemma-4 instruct bases. Two `decider-chat` repositories are stock models with a config
- Licence: Apache-2.0 for the code and the model repositories. Training uses about 95 public datasets under their own licences, which we didn't review
- Question types: noul (probability of yes), choice (2 to 255 options) and score (2 to 10 levels), any number a request, up to 1,024 scoring rows
- Context: 32k tokens per the README. The server cuts states at 32,768 tokens (`DECIDER_MAX_STATE_TOKENS`), and `/decide` caps context at 1,536
- Input: State as text or JSON. Images only with the separate decider-2b-vision model, which the README says is on older text weights
- Hardware: CUDA (about 4 GB for the 2B, 8.4 GB for the 4B, 65 GB for the 35B in bf16), Apple silicon (MPS, optional MLX kernel), CPU, and llama.cpp for the GGUF files. decider-31b needs vLLM and NVFP4
- Hosted option: None found
- Training: `scripts/train.sh full` reproduces the supervised stages of the 2B, 5.3 hours on a GH200 per the README. The RL stage and the mixture-v2 builders aren't in the package

- Scores: Reliability 90, Performance pending, Schema & documentation 78, Agent ergonomics 80, Security & auth 38, Payments & pricing 60, Task success pending, Maintenance & community 87, Transparency & trust 46 · total over the 7 assessed categories
- Why: Reliability, Scored on the local-package checklist, since Decider is open weights the owner runs, as for Kev and Strands Decider. · Schema & documentation, Read for a model you serve yourself. · Agent ergonomics, Read as an API an agent calls for a decision, as for the other decision models. · Security & auth, Read as software you run. · Payments & pricing, Free Apache-2.0 software with nothing to buy, so 20 + 20 + 20 for pricing, free use and no sign-up, by the self-hosted rule. · Maintenance & community, Read for an open-weight model. · Transparency & trust, Apache-2.0 for the code and model repositories, with the training code, data builders, teacher data and per-stage measurements published.
- Sources: 10, open questions: 7, both in the full twin
- Capabilities: inference.decision
- JSON: https://www.anchorterminal.com/api/v1/tools/decider.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/decider.svg` or a link to https://www.anchorterminal.com/tools/decider from a page under github.com/mapika, or the README of github.com/Mapika/decider, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.

## Before you call it

1. Pin weights by Hub tag (`v10`, `v2`) when results must repeat. The `main` branch of each model repository changes with new versions
2. Read `x_p_max` for the top probability. Since 1.3.0 `confidence` on choice and score answers follows TypeSafe's rescaled definition, not the top probability
3. Keep the server on 127.0.0.1 or put an authenticating proxy in front. It has no key option
4. Count state tokens before sending. Over 32,768 the state is cut silently, and `/decide` caps context at 1,536 tokens
5. Split multi-step arithmetic or multi-hop judgements into several questions, and don't write long rules into a question. The README says both fail

## Connect

```bash
pip install decider-ai
```

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| OpenAI Decisions API | BB | 71.5 | inference.decision | https://www.anchorterminal.com/tools/openai-decisions-api.min.md |
| Laya | B | 69.2 | inference.decision | https://www.anchorterminal.com/tools/convai-laya.min.md |
| Kev | B | 67.4 | inference.decision | https://www.anchorterminal.com/tools/jaredpalmer-kev.min.md |
| Vela 2.0 | B | 66.5 | inference.decision | https://www.anchorterminal.com/tools/vela.min.md |
| Clef | B | 66.1 | inference.decision | https://www.anchorterminal.com/tools/cloudflare-clef.min.md |


## Panel reviews (0, desk reviews from public material, no calls made)