Decision modelOpen weights
pplx-decider-v1.1-27b
Perplexity
- Released
- October 6, 2026
- Data date
- October 7, 2026
pplx-decider-v1.1-27b updates Perplexity’s first Decider model on the same Qwen3.8-27B base. Its separate readout head scores options as probabilities; this checkpoint does not include a full language-model head for free-form answers.
The weights have been publicly available under Apache 2.0 since October 6, 2026. Perplexity's announcement also states a hosted-use price of $0.02 per 1M input tokens.
Specifications and access
| Specification | Value and source |
|---|---|
| Model class | Multimodal decision model with a dedicated readout instead of a full text-generation headSource |
| Input modalities | Text and imagesSource |
| Output modalities | Scored answer options as probabilitiesSource |
| Checkpoint | perplexity-ai/pplx-decider-v1.1-27bSource |
| Base model | Qwen/Qwen3.8-27BSource |
| Parameters | 26,085,330,160 in Safetensors metadata; 27B is the named scaleSource |
| License | Apache-2.0; the public Hugging Face repository is not access-gatedSource |
Page 1 of 7
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
CV-Bench · public score (Show measurement, test conditions, and source)
- Source value
- 74.74
- Score
- 74.74
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
BLINK · public score (Show measurement, test conditions, and source)
- Source value
- 54.73
- Score
- 54.73
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
RealWorldQA · public score (Show measurement, test conditions, and source)
- Source value
- 57.83
- Score
- 57.83
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
CharXiv · public score (Show measurement, test conditions, and source)
- Source value
- 77.19
- Score
- 77.19
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
InfographicVQA · public score (Show measurement, test conditions, and source)
- Source value
- 79.78
- Score
- 79.78
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
Mind2Web · public score (Show measurement, test conditions, and source)
- Source value
- 68.76
- Score
- 68.76
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
Winoground · public score (Show measurement, test conditions, and source)
- Source value
- 84.5
- Score
- 84.5
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
KIE (CORD+FUNSD) · public score (Show measurement, test conditions, and source)
- Source value
- 91.32
- Score
- 91.32
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
Moderation (Hateful Memes) · public score (Show measurement, test conditions, and source)
- Source value
- 42.79
- Score
- 42.79
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
R-Bench-M · public score (Show measurement, test conditions, and source)
- Source value
- 29.25
- Score
- 29.25
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
MMMU-Pro vision · public score (Show measurement, test conditions, and source)
- Source value
- 28
- Score
- 28
- Metric
- public vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=public. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
CV-Bench · private score (Show measurement, test conditions, and source)
- Source value
- 41.71
- Score
- 41.71
- Metric
- private vision benchmark score
- Unit
- %
- Benchmark version
- Decision Index 0.3 vision suite
- Category
- decision-index-vision
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Jev Decision Index 0.3 vision-results data
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Jev Decision Index 0.3 vision evaluation of exact model https://huggingface.co/perplexity-ai/pplx-decider-v1.1-27b System 1; column=private. The evaluator reports this score on its vision suite. Exact image preprocessing and per-task sample sizes are not published in the JSON record.
- Context
- Ran through the checkpoint's own decision model, which accepts images.
Page 1 of 7
Published prices
- Price stated in the announcement
- Perplexity states $0.02 per 1M input tokens.Source