aiplans
Back to list
decisionavailableverified

Perplexity Decider V1.3 Flash

Perplexity

Decision tasks, including locate (text spans with offsets and scores) on v1.3.

Data basis
Input → Output
text, image → text
Pricing unit
1M tokens
Access
apiopen-weights
Last verified
2026-11-13

Pricing and verification

Official: $0.01/M input; output free. Vercel AI Gateway: same rate. OpenRouter :free: $0/$0

Per 1M input tokens
$0.01
Per 1M output tokens
$0
Per 1M tasks
—
Context
262144 tokens
Price status
Priced
Verified
2026-11-13

Perplexity Decisions API model pplx-decider-flash-v1.3 (4B). Official rate: $0.01 / 1M input tokens; output tokens free; no per-request fee. Accepts text and images; the new locate question type is text-only and does not work with web grounding. Alias pplx-decider-flash-latest (and perplexity/pplx-decider-flash-latest) selects this model and bills at its rate. v1.2 remains available under the pinned id. Official docs label the context window as 256K tokens; Vercel AI Gateway and OpenRouter publish 262,144 (256 × 1024). The catalog stores 262,144, same as v1.2. Web grounding adds Fast Search at $0.001 per search ($1 / 1,000) and $0.001 per successful URL fetch, on top of input tokens. Not a standalone search-API row. Apache-2.0 open weights on Hugging Face https://huggingface.co/perplexity-ai/pplx-decider-flash-v1.3 (cardData license apache-2.0). Vercel AI Gateway perplexity/pplx-decider-flash-v1.3: $0.01/1M input, max output tokens 0, context 262,144. OpenRouter pplx-decider-flash-v1.3:free is a published free channel (Input Free / Output Free, context 262K, released 2026-11-12). The public /api/v1/models list does not include it; paid OpenRouter pages for the non-free ids 404.

Model performance benchmarks
No audited, comparable public benchmark score is available yet. Music and world models do not currently have a unified public leaderboard; video models are seeded from public sources such as VBench / VBench++ where available.
Capabilities

Perplexity Decisions API model pplx-decider-flash-v1.3 (4B). Official rate: $0.01 / 1M input tokens; output tokens free; no per-request fee. Accepts text and images; the new locate question type is text-only and does not work with web grounding. Alias pplx-decider-flash-latest (and perplexity/pplx-decider-flash-latest) selects this model and bills at its rate. v1.2 remains available under the pinned id. Official docs label the context window as 256K tokens; Vercel AI Gateway and OpenRouter publish 262,144 (256 × 1024). The catalog stores 262,144, same as v1.2. Web grounding adds Fast Search at $0.001 per search ($1 / 1,000) and $0.001 per successful URL fetch, on top of input tokens. Not a standalone search-API row. Apache-2.0 open weights on Hugging Face https://huggingface.co/perplexity-ai/pplx-decider-flash-v1.3 (cardData license apache-2.0). Vercel AI Gateway perplexity/pplx-decider-flash-v1.3: $0.01/1M input, max output tokens 0, context 262,144. OpenRouter pplx-decider-flash-v1.3:free is a published free channel (Input Free / Output Free, context 262K, released 2026-11-12). The public /api/v1/models list does not include it; paid OpenRouter pages for the non-free ids 404.

Pricing and plans status

Official: $0.01/M input; output free. Vercel AI Gateway: same rate. OpenRouter :free: $0/$0

Note: multimodal pricing can be seconds, generations, credits, minutes, plan allowances or enterprise contracts. It is not forced into token pricing; only verified source-backed rows appear below.

Pricing sources and usage units

No structured pricing reference row yet; showing official pricing unit and source links above.

Related plans

No mapped plan yet. Next step is to materialize plans for Runway, Kling, Pika, Luma, Suno, Udio and similar providers.