Nimble 9B — VRAM & /v1/systemone setup
Written by Jakub Rusinowski · Last updated
Bespoke Labs' 9B decision model: a LoRA on Qwen3.5-9B. The default decision model in Ollama's library, called at /v1/systemone.
Nimble 9B needs about 7 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 Nimble 9B
Pull it with Ollama 0.35 or newer, then POST to /v1/systemone. Do not use ollama chat commands — this is not a chat model.
ollama pull nimblecurl http://localhost:11434/v1/systemone \
-H 'Content-Type: application/json' \
-d '{
"model": "nimble",
"state": {
"ticket": "I was charged twice. Please refund the extra payment."
},
"questions": {
"team": {
"type": "choice",
"instructions": "Which team should handle this ticket?",
"criteria": {
"billing": "Payments and refunds",
"technical": "Bugs and integrations",
"other": "None of the above"
}
}
}
}'Build a request for this model Decision models guide
Other decision models: 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
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 |
|---|---|---|---|
| Ollama build Publisher build · 9.5 GB file | 11.5 GB | Fits | |
| Q4_K_M 4.83 bpw · modelled quant | 7.9 GB | Fits | |
| Q6_K 6.56 bpw · modelled quant | 10 GB | Fits | |
| Q8_0 8.50 bpw · modelled quant | 12.3 GB | Fits |
Published file size: Ollama build 9.5 GB. A download size from the model publisher — not a VRAM requirement.
Ollama build is 9.5 GB. Which Bespoke checkpoint the tag packages (Sep 24 update vs original-2676) is not stated in one place; do not print a version.
How it was scored
Two different suites on two different scales. Never compare the numbers across the two cards.
Author's public benchmark
Bespoke Labs public benchmarks (13 datasets, 3,880 decisions)- choice questions
- 81.6%
- noul questions
- 80.2%
- score questions
- 54.6%
Decision Index 0.2.1 (snapshot 2026-09-28)
Decision Index 0.2.1- ECE (lower is better)
- 0.024
- Median compute
- 77.3 ms on 1x NVIDIA RTX PRO 6000 (96 GB)
Community-maintained; not affiliated with the model authors.
Source ↗The Index scored the repository bespokelabs/Bespoke-Nimble-9B-v2, not the checkpoint this page describes.
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
- 9.65B
- Context window
- 8K (prompt limit)
- Architecture
- Fine-tune of Qwen/Qwen3.5-9B
- Provider
- Bespoke Labs
- Licence
- Apache-2.0
- Specified at
- Q4_K_M
- System RAM
- 14 GB
- Record updated
- 2026-09-30
Commercial use permitted. No usage restrictions beyond attribution.
Nimble 9B — frequently asked questions
What is Nimble 9B?
Nimble 9B 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 Nimble 9B locally?
Install Ollama 0.35 or newer, run `ollama pull nimble`, then POST your state and questions to http://localhost:11434/v1/systemone. It is called through the API, not a chat session.
How much memory does Nimble 9B need?
About 7 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 Nimble 9B?
It has two separate published scores on two different suites — the author's own benchmark and the community Decision Index 0.2.1. They are not comparable with each other, and neither is a calibration guarantee.