pplx-decider is a 27B decision model from Perplexity, fine-tuned from Qwen3.8-27B. Give it a state (text, JSON, or images) and a set of typed questions, and it returns a calibrated probability for ev…
pplx-decider is a 27B decision model from Perplexity, fine-tuned from Qwen3.8-27B. Give it a state (text, JSON, or images) and a set of typed questions, and it returns a calibrated probability for every answer in a single forward pass. It doesn't generate text.
It's compatible with the Jev and System One APIs, so it works through Ollama's `/v1/systemone` endpoint and `ollama run --ask`.
```
ollama pull pplx-decider
```
## Highlights
- **Accurate**: 85.7% across 11 public benchmarks in Perplexity's evaluation, ahead of Jev (84.5%) and its Qwen3.8-27B base (74.8%).
- **Multimodal**: decide over text, JSON, or images such as screenshots, receipts, and forms.
- **One forward pass**: every option of every question is scored together, with no output tokens.
- **Smaller option**: [pplx-decider-flash](https://ollama.com/library/pplx-decider-flash) is the 4B version — the same request format and image support, with most of this model's accuracy, at about one seventh of the size.
- **Apache 2.0 license**
## What you can build
| Task | You define | You get back |
|---|---|---|
| Route a request | The destinations and when each applies | The chosen destination and each destination's probability |
| Check a condition | A yes or no question | True or false, and the probability of each |
| Grade an answer | A rubric of levels | The probability-weighted level and each level's probability |
| Gate an agent step | A condition and the agent's proposed action | Go or no-go, with the probability of each |
## CLI
```
ollama run pplx-decider --ask "Is a refund being requested?" < ticket.txt
```
```
ollama run pplx-decider --ask "Which team should handle this?" \
--choice billing="Payments and refunds" \
--choice technical="Outages and errors" < ticket.txt
```
```
ollama run pplx-decider --ask "How urgent is this?" \
--level "Routine" \
--level "Soon" \
--level "Immediate" < ticket.txt
```
The state can also follow the model name instead of coming from stdin.
## API
Decision models use Ollama's `/v1/systemone` endpoint, [documented here](https://docs.ollama.com/capabilities/decision). Put the text you want judged in `state` and your questions in `questions`. Ollama builds pplx-decider's prompt for you.
| Field | Description |
|---|---|
| `model` | `pplx-decider` |
| `state` | The text to judge. Use a string, or a JSON object or array for structured input. |
| `images` | Optional. Base64-encoded PNG, JPEG, or WebP images shared by all questions, in request order. |
| `questions` | 1 to 64 named questions. Answers come back in the same order. |
| `keep_alive` | Optional. How long the model stays loaded after the request. |
Every question has a `type`, `instructions`, and usually `criteria`:
| Type | `criteria` | Answer fields |
|---|---|---|
| `choice` | An object mapping each option to a description. Use `null` to let the option name describe itself. | `choice`, `probabilities`, `confidence` |
| `noul` | Optional. `{"true": "...", "false": "..."}` to describe each side. | `noul`, the probability that the answer is true |
| `score` | An array of 2 to 10 level descriptions, lowest first. Levels are numbered from 0. | `score` (the probability-weighted level), `legend`, `probabilities`, `confidence` |
`confidence` runs from 0 to 1 and shows how strongly the probabilities favor one answer. It isn't the chance that the answer is right.
### cURL
```
curl http://localhost:11434/v1/systemone -d '{
"model": "pplx-decider",
"state": "Our app crashes on launch for every user since this morning'\''s update. Please fix it now.",
"questions": {
"urgent": {"type": "noul", "instructions": "Is this support request urgent?"},
"team": {
"type": "choice",
"instructions": "Which team should handle this request?",
"criteria": {
"billing": "Payments, invoices, and refunds",
"technical": "Outages, errors, and configuration",
"sales": "Plans and upgrades"
}
},
"severity": {
"type": "score",
"instructions": "How severe is the customer impact?",
"criteria": ["No impact", "Minor", "Major", "Critical"]
}
}
}'
```
Response:
```json
{
"model": "pplx-decider",
"answers": {
"urgent": {"type": "noul", "noul": 0.994},
"team": {
"type": "choice",
"choice": "technical",
"probabilities": {"billing": 0.002, "technical": 0.996, "sales": 0.002},
"confidence": 0.994
},
"severity": {
"type": "score",
"score": 2.975,
"legend": {"0": "No impact", "1": "Minor", "2": "Major", "3": "Critical"},
"probabilities": {"0": 0.001, "1": 0.002, "2": 0.017, "3": 0.98},
"confidence": 0.975
}
},
"usage": {"input_tokens": 338, "output_tokens": 0}
}
```
### Python
Use TypeSafe's official Python SDK and point it at Ollama. The SDK requires an API key, but Ollama ignores it, so any value works.
```
pip install typesafe-sdk
export TYPESAFE_BASE_URL=http://localhost:11434
export TYPESAFE_API_KEY=ollama
export TYPESAFE_DEFAULT_MODEL=pplx-decider
```
```python
from typesafe_sdk import Choice, Noul, Score, TypeSafeClient
questions = {
"urgent": Noul(instructions="Is this support request urgent?"),
"team": Choice(
instructions="Which team should handle this request?",
criteria={"billing": "Payments, invoices, and refunds", "technical": "Outages, errors, and configuration", "sales": "Plans and upgrades"},
),
"severity": Score(instructions="How severe is the customer impact?", criteria=["No impact", "Minor", "Major", "Critical"]),
}
with TypeSafeClient(timeout=120) as client:
result = client.system_one(state="Our app crashes on launch for every user since this morning's update. Please fix it now.", questions=questions)
print(result.choices["team"].choice) # technical
```
### Images
```
curl http://localhost:11434/v1/systemone -d '{
"model": "pplx-decider",
"state": "The agent wants to submit the attached form.",
"images": ["<base64-encoded image>"],
"questions": {
"complete": {"type": "noul", "instructions": "Are all required fields filled in?"},
"kind": {
"type": "choice",
"instructions": "What kind of document is this?",
"criteria": {"invoice": null, "receipt": null, "contract": null, "other": null}
}
}
}'
```
## Benchmarks
Accuracy reported by Perplexity, measured through the Perplexity API. See the [model card](https://huggingface.co/perplexity-ai/pplx-decider-v1-27b) for all 11 benchmarks.
| Benchmark | pplx-decider | Jev | Qwen3.8-27B |
|---|---:|---:|---:|
| RAGTruth | **88.8%** | 77.3% | 61.5% |
| FinancialPhraseBank | **84.2%** | 77.0% | 75.7% |
| TabFact | **90.6%** | 89.8% | 78.6% |
| JudgeBench | 78.3% | **78.6%** | 68.9% |
| WinoGrande | 83.3% | **90.7%** | 73.1% |
| **Overall (11 benchmarks)** | **85.7%** | 84.5% | 74.8% |
## License
[Apache 2.0](https://huggingface.co/perplexity-ai/pplx-decider-v1-27b/blob/main/LICENSE)