Perplexity · Official-source reference
Perplexity Decider Flash V1.2
Official pricing
$0.01 / $0
Perplexity · input / output per million tokens · Perplexity charges $0.01 per million input tokens, with free output. A shared-state request reads and bills its state once; splitting questions repeats the state cost. This differs from services that bill a separate prompt for every question.
Context
262.1K
Perplexity Decider Flash V1.2 · Specifications
| property | value |
|---|---|
| Parameters | 4B |
| Context | 262,144 tokens; validate long-context accuracy on your task |
| Checkpoint license | Apache-2.0 (HF model card) |
| State input | Text, JSON and images |
| Answer types | noul / choice / score; typed probabilities |
| Moving alias | pplx-decider-flash-latest |
Perplexity Decider Flash V1.2 · Weight license
| Model weights | License |
|---|---|
| Open weights | Apache 2.0 |
Perplexity Decider Flash V1.2 · Peer comparison
| model | score | input | output |
|---|---|---|---|
| Decider V1.2 | HF Full 65.5 | $0.02 | $0 |
| Decider Flash V1.2 | HF Full 54.8 | $0.01 | $0 |
| d1-ultra | AA 65; not comparable to HF | $0.042 | $0 |
| Jev 1.14 | HF Full 61.5; AA 57 | $0.042 | $0 |
Perplexity Decider Flash V1.2 · Cost examples
| workload | tokens | usd |
|---|---|---|
| 1 million input tokens | 1,000,000 | $0.01 |
| 10 million input tokens | 10,000,000 | $0.10 |
| 1,000 requests × 1,000 input tokens | 1,000,000 | $0.01 |
Perplexity Decider Flash V1.2 · Benchmarks
| metric | value | scope |
|---|---|---|
| HF Full score | 54.8 | HF 0.3.2; public + private |
| Public suite, joint requests | 58.88 | Publisher model card; suite 0.3 |
| Public suite, one question per request | 60.05 | Publisher model card; suite 0.3 |
| Vision public suite | 72.53 | Publisher model card; vision suite 0.3 |
Cost examples
Flash recommends one question per request by default. Group closely related routing steps or select-all-that-apply decisions only after testing. Independent documents, tools or candidate calls should be scored separately.
Context and limits
Perplexity charges $0.01 per million input tokens, with free output. A shared-state request reads and bills its state once; splitting questions repeats the state cost. This differs from services that bill a separate prompt for every question.
Questions about Perplexity Decider Flash V1.2 pricing
Perplexity charges $0.01 per million input tokens, with free output. A shared-state request reads and bills its state once; splitting questions repeats the state cost. This differs from services that bill a separate prompt for every question.
Call POST https://api.perplexity.ai/v1/decisions with model pplx-decider-flash-v1.2. OpenRouter also serves the model through its Decisions API, not chat completions; Vercel AI Gateway lists it. The API allows 128 questions per request, 255 choice options and 10 rubric levels.
Compare official token rates below. HF Full scores and AA Decision Index scores use different benchmarks and cannot be ranked together. Test each model on the same labeled examples and thresholds before choosing.
No. pplx-decider-flash-latest follows new releases in its own model family. Perplexity routes legacy 27B IDs pplx-decider-v1.1-27b and pplx-decider-v1-27b to Decider V1.2 (27B). Revalidate probability thresholds when the underlying version changes.