OpenDecider-small MLX 4-bit — VRAM & Python setup
Written by Jakub Rusinowski · Last updated
OpenDecider-small quantised to 4-bit for Apple Silicon with MLX: a 2.3 GB download and about 2.6 GB of memory, roughly 2 points below full precision on typed-decisions.
OpenDecider-small MLX 4-bit needs about 3 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
Call OpenDecider-small MLX 4-bit
A Python library: you load it inside your own program. It has no HTTP server and does not run in Ollama.
- Question types
- choice noul (yes / no) score
- Inputs
- text
- Serve with
opendecider (MLX)
pip install "opendecider[mlx]"Not in the request builder: this build is documented for the Python library only.
Related guides: Confidence, thresholds and calibration
Other decision models: Nimble 9B · Tev1 4B · Tev1 0.8B · Winnow 12B · Winnow E4B · Decider 2B · Decider 4B · Decider 35B-A3B (NVFP4) · JevK5 4B · Intern-Decision 4B · AutoJev 27B · Laya · Clef 27B · Clef-Flash 9B · Rune 26B-A4B v3 · OpenDecider-nano · OpenDecider-small · Bosun v3.1 0.6B · Bosun v3.1 1.7B · d1-3B · d1-omni-600M · THX-01 322M
Hardware fit
Weights plus overhead plus the KV cache at a 8,192-token prompt, on NVIDIA RTX 4090 (24 GB). A publisher build is sized from its file; the other rows are modelled at a standard quant. Decision requests are short, so no long-context figure is shown.
| Quant | Memory | VRAM | Fit |
|---|---|---|---|
| MLX 4-bit Publisher build · 2.26 GB file | 4 GB | Fits | |
| Q4_K_M 4.83 bpw · modelled quant | 4.1 GB | Fits | |
| Q6_K 6.56 bpw · modelled quant | 5 GB | Fits | |
| Q8_0 8.50 bpw · modelled quant | 6 GB | Fits |
Published file size: MLX 4-bit 2.26 GB. A download size from the model publisher — not a VRAM requirement.
Apple Silicon only. About 2 points below the full-precision model on typed-decisions (0.651 vs 0.672). Serving this build over HTTP is not documented, so it is listed as a Python library.
How it was scored
Each table is one suite. Numbers in different tables were measured on different questions and are not comparable, with each other or with the Decision Index.
Decision Index 0.2.1 (snapshot 2026-09-28)
Decision Index 0.2.1Community Decision Index (not by the model's authors)OpenDecider is not on the Decision Index 0.2.1 snapshot (28 September 2026), and its author has not published an Index score. Community-maintained; not affiliated with the model authors.
Source ↗| What was scored | Score | Source |
|---|---|---|
| Test split | 65.1% | Source ↗ |
Zero-shot. The full-precision model scores 0.672 here.
| What was scored | Score | Source |
|---|---|---|
| 200 decisions | 74.5% | Source ↗ |
Scored by the authors with the public Jev-vs-Laya harness (https://github.com/pavanjava/jev_and_laya_benchmarking). ECE is measured on this suite, not on typed-decisions.
Published latency
| Device | Time | Condition | Who measured it | Source |
|---|---|---|---|---|
| Apple M4 Max, 64 GB | 65 ms | 1 short question | Authors' own internal benchmark — not independently reproduced | Source ↗ |
| Apple M4 Max, 64 GB | 143 ms | median over the benchmark questions | Authors' own internal benchmark — not independently reproduced | Source ↗ |
As published, on the device and under the condition shown. A decision model answers in one forward pass, so this is time per decision, not tokens per second.
Decision models are not ranked on chat, creative or coding scores.
Specifications
Verified — Checked against the primary source — the model card or the vendor spec page — and corroborated by a second independent source.
- Parameters
- 4.02B
- Context window
- Not published
- Architecture
- Fine-tune of Qwen/Qwen3-4B-Instruct-2507
- Provider
- Manjunath Janardhan
- Licence
- Apache-2.0
- Availability
- Weights downloadable
- Practical deployment
- Edge device (phone, SBC)
- Specified at
- Q4_K_M
- System RAM
- 8 GB
- Record updated
- 2026-10-13
- Runs with
- MLX: supported
Commercial use permitted. No usage restrictions beyond attribution.
Other OpenDecider sizes
OpenDecider-small MLX 4-bit — frequently asked questions
What is OpenDecider-small MLX 4-bit?
OpenDecider-small MLX 4-bit is a decision model: you send it a state and typed questions (choice, yes/no/unknown, or a score) and it returns one answer per question with a probability for every option. It is not a chat model.
How do I run OpenDecider-small MLX 4-bit locally?
A Python library: you load it inside your own program. It has no HTTP server and does not run in Ollama. Install: `pip install "opendecider[mlx]"`.
How much memory does OpenDecider-small MLX 4-bit need?
About 3 GB for the weights plus overhead at Q4_K_M, before the prompt's KV cache. Decision prompts are short, so the cache stays small.
How accurate is OpenDecider-small MLX 4-bit?
It has no community Decision Index score. Every score on this page comes from its authors, on the suites named beside each table; none is an independent result.