Perplexity pricing splits into two separate systems: Pro/Max/Enterprise are subscriptions that trade a monthly or yearly fee for search and research quota, while the Sonar API is pay-as-you-go, charging per token and per request — the two paths are fully independent and their money does not mix. On the subscription side, Pro starts at $20/mo and Max at $200/mo; on the API side, sonar starts at $1/$1 per 1M tokens plus $5–$12 per 1K request fees. All prices below are as of the official pages on 2026-08-18.
Perplexity Pricing Comparison: Subscriptions and the Sonar API Are Two Separate Systems
You cannot look up a single number for Perplexity pricing — the consumer/enterprise subscriptions and the Sonar API are two fully independent billing systems, and subscription quota does not equal API balance. Subscriptions are activated at perplexity.ai/pro, paying monthly or yearly for functional quota like Pro Search, Deep Research, and Computer; the Sonar API is billed per token and request on the official pricing docs, with prepaid balance deducted as you call. Here is the high-level comparison (subscription pricing, API pricing):
| Dimension | Perplexity Subscription | Perplexity Sonar API |
|---|---|---|
| Billing model | Subscription; monthly/yearly for quota | Pay-as-you-go; token fees + request fees |
| Official entry | perplexity.ai/pro | API docs / settings/api |
| Top-up | Subscription charges; annual available | Prepaid balance in the console |
| Models | Multi-model switching: GPT, Claude, Gemini, Kimi, Sonar 2 | Four priced Perplexity-owned Sonar models |
| Entry price | Pro $20/mo | sonar $1/$1 per 1M tokens |
| Rate limits | Quota (Pro Search / Deep Research counts) | Official rate limits, by account tier |
| Best for | Search, research, reports in Web/App | Building apps, product integration, server-side calls |
Subscriptions are a quota-pool model: you buy functional allowance that resets monthly at a fixed price. The API is a balance model: every call deducts by token count and request count, so cost grows linearly with usage. The clearest difference: subscription fees are fixed and cover what you need; the API charges exactly what you use. Also note that the Sonar API covers only Perplexity-owned grounded search models — third-party models hosted via the Agent API bill separately.
Perplexity Subscription Pricing: Choosing Among Free, Pro, Max, and Enterprise Tiers
Perplexity offers five subscription tiers: Free, Pro at $20/mo ($200/yr), Max at $200/mo ($2,000/yr), Enterprise Pro at $40/seat/mo, and Enterprise Max at $325/seat/mo, with annual billing saving about 16% across the board. Tier pricing and quota (source):
| Tier | Monthly | Annual | Pro Search | Deep Research | Highlights |
|---|---|---|---|---|---|
| Free | Free | — | ~200/wk | ~20/mo | Basic search workflow, limited file uploads |
| Pro | $20/mo | $200/yr (~$16.7/mo) | Unlimited | ~2.5× Free | Model switching + Premium Sources (Statista, PitchBook) + Labs |
| Max | $200/mo | $2,000/yr (~$166.7/mo) | ~20× Free | ~25× Free | o3-pro, Claude Opus 4.6/4.8, Model Council, unlimited Computer, ~10,000 Computer credits/mo |
| Enterprise Pro | $40/seat/mo | $400/seat/yr (~$33.3/mo) | 400/wk/user | 50/mo/user | SSO/SCIM, internal knowledge search (up to 500 files), Zero Data Training |
| Enterprise Max | $325/seat/mo | $3,250/seat/yr (~$270.8/mo) | 4,000/wk/user | 500/mo/user | Unlimited Computer, expanded audit logs, dedicated account manager |
The most misunderstood relationship is Pro vs Max: Max costs 10× Pro, but what it adds is Max-exclusive capability — o3-pro, Claude Opus, Model Council, unlimited Computer, and ~10,000 Computer credits/month — not simply 10× the quota. Pro's model switching spans OpenAI's GPT-5.6 Terra, Claude's Claude Sonnet 5, Gemini's Gemini 3.1 Pro, Kimi's Kimi K3, and Perplexity's own Sonar 2, making cross-vendor answer comparison its core use case. How to pick: Pro is enough for daily search and research; Max pays off only for o3-pro/Opus-level analysis or heavy Computer deliverables; teams needing SSO and internal knowledge search start with Enterprise Pro and step up to Enterprise Max at scale; start with Free if you are unsure whether the product fits your research habits.
Perplexity Sonar API Pricing Comparison: Token Rates and Request Fees
The Sonar API bill has two parts — token fees and request fees: sonar starts at $1/$1 per 1M tokens with request fees of $5/$8/$12 per 1K by Low/Medium/High search context, and the Pro, reasoning, and Deep Research models cost more. Model pricing (source):
| Model | Input | Output | Request fee (per 1K) | Notes |
|---|---|---|---|---|
| sonar | $1/M | $1/M | Low/Med/High = $5/$8/$12 | Basic live search, cost-sensitive pick |
| sonar-pro | $3/M | $15/M | $6/$10/$14; Pro Search = $14/$18/$22 | ~2× search results, multi-step Pro Search |
| sonar-reasoning-pro | $2/M | $8/M | $6/$10/$14 | Chain-of-thought reasoning + live search, replacing retired sonar-reasoning |
| sonar-deep-research | $2/M | $8/M | + Citation $2/M, Search $5/1K, Reasoning $3/M | Exhaustive research across hundreds of sources |
sonar-pro's token rates are 3× (input) and 15× (output) the base sonar, but it returns roughly 2× the search results and supports Pro Search (enable search_type=pro with stream=true in web_search_options); Pro Search request fees run noticeably higher than standard tiers ($14/$18/$22 vs $6/$10/$14). The most overlooked cost on the API side is request fees: the three search-context levels are priced differently, so lightweight retrieval on Low keeps costs down while research queries justify High; Deep Research layers on Citation, Search Queries, and Reasoning charges. The Search API (raw web search at $5/1K with no token fees) and Embeddings API (from $0.004/M) are separate product lines outside Sonar chat model billing — do not mix them into the same cost estimate.
Quota, Usage and Rate Limits: Subscription Allowances vs API Metering
Subscriptions are about a "quota pool," while the API is about "balance + rate limits" — two completely different consumption and rate-limiting models. On the subscription side, Pro offers unlimited personal Pro Search, Free runs ~200/week, Enterprise grants 400/4,000 per week per user with Deep Research at 50/500 per month, and Max delivers unlimited Computer plus ~10,000 Computer credits/month with quota resetting monthly. On the API side, official rate limits vary by account tier, every call deducts balance by token and request count, and running out of balance or hitting limits breaks production calls.
Do not look only at unit price before launch: for API workloads, confirm your account tier can carry target traffic and the balance is sufficient; for subscriptions, judge whether the quota fits — users treating Perplexity as a daily production tool with continuous Computer tasks need Max's unlimited Computer and ~10,000 credits, while Free/Pro allowances suit intermittent research. The consumer model lineup also evolves with releases, with o3-pro and Claude Opus 4.6/4.8 currently held mostly in the Max tier.
Top-Up, Deals, Official Entry and API Key Setup
There are two official entries: subscriptions are activated at perplexity.ai/pro with annual billing saving about 16%, while the API is topped up in the settings/api console and billed per token and request — API keys are issued in the same console. Subscriptions are charge-on-signup with monthly or annual billing: Pro $200/yr, Max $2,000/yr, Enterprise Pro $400/seat/yr, Enterprise Max $3,250/seat/yr, all around 16% off; annual Max billing is mainly available on the web, and mobile checkout may differ (subscription pricing).
Know the product-line boundaries on the API side: the Sonar API covers only Perplexity-owned Sonar models; the Agent API (hosting third-party GPT/Claude/Gemini models) bills at direct provider token rates, and the Search API and Embeddings API are priced independently. Consumer subscriptions and the API are separate systems: Pro/Max quota cannot be used as API balance, though some Pro accounts include small API credits per the account view. For integration, the Sonar API supports REST and OpenAI SDK-compatible access — Claude Code, Cline, and similar tools just configure an OpenAI-compatible base URL and model name such as sonar or sonar-pro (API pricing).
Is It Worth It: How to Buy Perplexity Smartly
Bottom line: Pro ($20/mo) is the best value for individual search and research; Max ($200/mo) is only worth it for power users living on o3-pro, Opus, and Computer; teams go Enterprise for a unified entry; and the Sonar API pay-as-you-go is the only path for building your own app. Subscription and API decisions follow different logic — a subscription trades a monthly budget for fixed quota, while the API trades per-use fees for flexibility — so pick the right entry point before comparing prices.
| Your situation | Recommendation |
|---|---|
| Occasional search, just trying it out | Free tier (limited Pro Search, ~20 Deep Research/mo) |
| Daily research, comparing answers across models | Pro $20/mo (annual $200, ~16% off) |
| Relying on o3-pro / Opus / Computer deliverables | Max $200/mo (annual $2,000) |
| Team needs SSO and internal knowledge search | Enterprise Pro from $40/seat/mo |
| Building apps, product integration, server-side calls | Sonar API pay-as-you-go (from $1/$1 + request fees) |
A few reminders: 1. subscription and API money do not mix, so choose the entry first; 2. the real cost of the Sonar API includes request fees, not just token rates; 3. the model lineup evolves with releases, and the Max-exclusive o3-pro and Claude Opus are the main tier differentiator; 4. Enterprise plans differ significantly on data training policy (contractual zero data training). For broader comparisons across vendors, check the DeepSeek API pricing page.