Perplexity API Pricing (Oct 2026): All 20 Models per 1M Tokens
As of : 20 models tracked. The median price is $0.70 per 1M input tokens and $1.60 per 1M output tokens. Cheapest blended price: pplx-7b-online (Perplexity, — per 1M input / $0.28 per 1M output).
API list prices for every Perplexity chat model we track.
Models
| Model | Provider | Input / 1M | Output / 1M | Cached input / 1M | Context |
|---|---|---|---|---|---|
| pplx-7b-online | Perplexity | — | $0.28 | — | 4K |
| sonar-small-online | Perplexity | — | $0.28 | — | 12K |
| mistral-7b-instruct | Perplexity | $0.07 | $0.28 | — | 4K |
| mixtral-8x7b-instruct | Perplexity | $0.07 | $0.28 | — | 4K |
| pplx-7b-chat | Perplexity | $0.07 | $0.28 | — | 8K |
| sonar-small-chat | Perplexity | $0.07 | $0.28 | — | 16K |
| llama-3.1-8b-instruct | Perplexity | $0.20 | $0.20 | — | 131K |
| sonar-medium-online | Perplexity | — | $1.80 | — | 12K |
| codellama-34b-instruct | Perplexity | $0.35 | $1.40 | — | 16K |
| pplx-70b-online | Perplexity | — | $2.80 | — | 4K |
| sonar-medium-chat | Perplexity | $0.60 | $1.80 | — | 16K |
| llama-3.1-70b-instruct | Perplexity | $1.00 | $1.00 | — | 131K |
| sonar | Perplexity | $1.00 | $1.00 | — | 128K |
| codellama-70b-instruct | Perplexity | $0.70 | $2.80 | — | 16K |
| llama-2-70b-chat | Perplexity | $0.70 | $2.80 | — | 4K |
| pplx-70b-chat | Perplexity | $0.70 | $2.80 | — | 4K |
| sonar-reasoning | Perplexity | $1.00 | $5.00 | — | 128K |
| sonar-deep-research | Perplexity | $2.00 | $8.00 | — | 128K |
| sonar-reasoning-pro | Perplexity | $2.00 | $8.00 | — | 128K |
| sonar-pro | Perplexity | $3.00 | $15.00 | — | 200K |
FAQ
How are prices shown?
In US dollars per 1 million tokens, as listed for each model's API. 'Blended' assumes 3 input tokens for every output token.
Which model is cheapest here?
pplx-7b-online from Perplexity, at — per 1M input and $0.28 per 1M output tokens (as of ).
How current are these prices?
We copy the open-source LiteLLM price list once a day and record every change. Providers can change prices at any time, so confirm on the official pricing page linked from each model.