# Sonar Reasoning Pro

Sonar Reasoning Pro is the top Sonar tier. It pairs the same deep multi-source web retrieval as Sonar Pro with the strongest reasoning in the lineup for hard research and analysis questions in a single API call.

- **Model ID:** `perplexity/sonar-reasoning-pro`
- **Type:** chat
- **Providers:** perplexity
- **Context window:** 127,000
- **Maximum output tokens:** 8,000
- **Pricing:** See the provider details below
- **Canonical page:** https://vercel.com/ai-gateway/models/sonar-reasoning-pro

## Supported parameters

Detailed capability metadata has not been reported for this model.

## Example

```ts
import { streamText } from 'ai'

const result = streamText({
  model: 'perplexity/sonar-reasoning-pro',
  prompt: 'Why is the sky blue?'
})
```

## About

Sonar Reasoning Pro is the full search-augmented reasoning stack from Perplexity. It combines the deep, multi-source web retrieval of Sonar Pro with chain-of-thought reasoning, producing answers that are both broadly sourced and logically structured.

Perplexity describes benchmark results and search modes for this model in its announcement. See https://sonar.perplexity.ai for methodology, pricing context, and product details.

The practical impact shows up on hard research questions. Queries that require reconciling conflicting sources, reasoning about implications of recent developments, or building a structured argument from scattered web evidence use both retrieval breadth and reasoning depth. The model searches more sources than Sonar Reasoning, applies more deliberation, and returns longer, more structured responses. The context window of 127K tokens supports extended research sessions where accumulated context from prior turns informs later searches and reasoning.

## What to consider

Sonar Reasoning Pro has the highest per-token cost and longest response times in the Sonar family. Use AI Gateway's cost tracking and latency monitoring to profile requests. Confirm the quality gain matches your budget. Web search calls bill separately (N/A per thousand when listed for this model).

## When to use

### Best For

- **Deep research and analysis** requiring both broad source retrieval and multi-step logical reasoning over the evidence
- **Strategic decision support** where analysis quality depends on both the breadth of sources consulted and the rigor of reasoning applied
- **Complex competitive intelligence** that needs to reconcile conflicting information from multiple sources into one assessment
- **Policy and regulatory analysis** requiring structured reasoning across evolving standards, guidelines, and jurisdictional differences
- **High-stakes factual analysis** where thoroughness of both retrieval and reasoning affects decision quality

### Consider Alternatives When

- The query is straightforward and doesn't need deep reasoning. Base Sonar or Sonar Pro answer faster at lower cost
- Latency is a hard constraint. Sonar Reasoning Pro's combined deep search and reasoning produces the longest response times in the Sonar family
- Budget per request is tightly constrained. Broad retrieval and extended reasoning generate the most output tokens
- The task doesn't benefit from web grounding. A standalone reasoning model fits better for closed-domain problems

## Best for

- **Deep research and analysis:** Requires both broad source retrieval and multi-step logical reasoning over the evidence
- **Strategic decision support:** Analysis quality depends on both the breadth of sources consulted and the rigor of reasoning applied
- **Complex competitive intelligence:** Needs to reconcile conflicting information from multiple sources into one assessment
- **Policy and regulatory analysis:** Requires structured reasoning across evolving standards, guidelines, and jurisdictional differences
- **High-stakes factual analysis:** Thoroughness of both retrieval and reasoning affects decision quality

## Consider alternatives

- **Straightforward queries:** Base Sonar or Sonar Pro answer faster at lower cost without deep reasoning
- **Hard latency constraints:** Combined deep search and reasoning produces the longest response times in the Sonar family
- **Tight per-request budget:** Broad retrieval and extended reasoning generate the most output tokens
- **Closed-domain problems:** A standalone reasoning model fits better when no web grounding is needed

## Frequently asked questions

### What makes Sonar Reasoning Pro different from Sonar Reasoning?

Sonar Reasoning Pro combines Pro-level deep web retrieval (more sources) with chain-of-thought reasoning. Standard Sonar Reasoning uses base-level retrieval with reasoning. The Pro variant goes further on both search and deliberation.

### How does Sonar Reasoning Pro compare to chaining a search API with a separate reasoning model?

Sonar Reasoning Pro integrates search, retrieval, source evaluation, reasoning, and cited answer generation into a single API call. This cuts latency and operational complexity versus wiring separate services. The model is tuned to reason over its own search results.

### What types of queries benefit most from Sonar Reasoning Pro?

Competitive intelligence, regulatory research, technical due diligence, and strategic analysis when the answer depends on reasoning across many sources. It fits questions that need both broad retrieval and structured analysis.

### What is the context window for Sonar Reasoning Pro?

127K tokens. This supports extended multi-turn research sessions where prior search results and reasoning inform subsequent queries.

### How much does Sonar Reasoning Pro cost?

Pricing appears on this page and updates as providers adjust their rates. AI Gateway routes traffic through the configured provider.

### How do I authenticate with Sonar Reasoning Pro through AI Gateway?

Use your AI Gateway API key with the model identifier ``perplexity/sonar-reasoning-pro``. AI Gateway handles provider routing and authentication. You don't need a separate Perplexity API key when using gateway-managed access.

### Does Sonar Reasoning Pro show its reasoning process?

Yes. The chain-of-thought reasoning trace appears in the response. It shows how the model analyzed and connected information from the retrieved sources to reach its conclusion.

## Links

- [Model page](https://vercel.com/ai-gateway/models/sonar-reasoning-pro)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://sonar.perplexity.ai)
- [Provider pricing](https://docs.perplexity.ai/guides/pricing)


If you encounter incorrect information, can’t find what you need, need a capability Vercel doesn’t currently offer, or have any other feedback to report, please do it by following the instructions at https://vercel.com/feedback.md.