> ## Documentation Index > Fetch the complete documentation index at: https://docs.perplexity.ai/llms.txt > Use this file to discover all available pages before exploring further. # Models > Explore available presets and third-party models for the Agent API, including Perplexity presets and third-party model support. ## Available Models The Agent API supports direct access to models from multiple providers. All models are accessed directly from first-party providers with transparent token-based pricing. Pricing rates are updated monthly and **reflect direct first-party provider pricing with no markup**. All charges are based on actual token consumption, and every API response includes exact token counts so you know your costs per request. Looking for pre-configured model setups? See [**Presets**](/docs/agent-api/presets) — optimized for specific use cases. Sonar — Perplexity's grounded search model. | Model | Input (\$/1M) | Output (\$/1M) | Cache (\$/1M) | Docs | | ------------------ | ------------- | -------------- | ------------- | ----------------------------------------------------------- | | `perplexity/sonar` | 0.25 | 2.50 | 0.0625 | [Sonar](https://docs.perplexity.ai/docs/sonar/models/sonar) | Claude Opus (highest reasoning), Sonnet (balanced), and Haiku (fastest, cheapest). | Model | Input (\$/1M) | Output (\$/1M) | Cache (\$/1M) | Docs | | ----------------------------- | ------------- | -------------- | ------------- | --------------------------------------------------------------------- | | `anthropic/claude-opus-4-7` | 5 | 25 | 0.50 | [Claude Opus 4.7](https://www.anthropic.com/news/claude-opus-4-7) | | `anthropic/claude-opus-4-6` | 5 | 25 | 0.50 | [Claude Opus 4.6](https://www.anthropic.com/news/claude-opus-4-6) | | `anthropic/claude-opus-4-5` | 5 | 25 | 0.50 | [Claude Opus 4.5](https://www.anthropic.com/news/claude-opus-4-5) | | `anthropic/claude-sonnet-4-6` | 3 | 15 | 0.30 | [Claude Sonnet 4.6](https://www.anthropic.com/news/claude-sonnet-4-6) | | `anthropic/claude-sonnet-4-5` | 3 | 15 | 0.30 | [Claude Sonnet 4.5](https://www.anthropic.com/news/claude-sonnet-4-5) | | `anthropic/claude-haiku-4-5` | 1 | 5 | 0.10 | [Claude Haiku 4.5](https://www.anthropic.com/news/claude-haiku-4-5) | GPT-5 family — flagship, mini, and nano variants. | Model | Input (\$/1M) | Output (\$/1M) | Cache (\$/1M) | Docs | | --------------------- | ------------- | -------------- | ------------- | -------------------------------------------------------------------- | | `openai/gpt-5.5` | 5.00 | 30.00 | 0.50 | [GPT-5.5](https://developers.openai.com/api/docs/models/gpt-5.5) | | `openai/gpt-5.4` | 2.50 | 15.00 | 0.25 | [GPT-5.4](https://platform.openai.com/docs/models/gpt-5.4) | | `openai/gpt-5.4-mini` | 0.75 | 4.50 | 0 | [GPT-5.4 Mini](https://platform.openai.com/docs/models/gpt-5.4-mini) | | `openai/gpt-5.4-nano` | 0.20 | 1.25 | 0 | [GPT-5.4 Nano](https://platform.openai.com/docs/models/gpt-5.4-nano) | | `openai/gpt-5.2` | 1.75 | 14 | 0.175 | [GPT-5.2](https://platform.openai.com/docs/models/gpt-5.2) | | `openai/gpt-5.1` | 1.25 | 10 | 0.125 | [GPT-5.1](https://platform.openai.com/docs/models/gpt-5.1) | | `openai/gpt-5` | 1.25 | 10 | 0.125 | [GPT-5](https://platform.openai.com/docs/models/gpt-5) | | `openai/gpt-5-mini` | 0.25 | 2 | 0.025 | [GPT-5 Mini](https://platform.openai.com/docs/models/gpt-5-mini) | Gemini 3 family — Pro for long-context, Flash and Flash Lite for speed. | Model | Input (\$/1M) | Output (\$/1M) | Cache (\$/1M) | Docs | | -------------------------------------- | ------------------------------ | -------------------------------- | ------------- | ----------------------------------------------------------------------------------------------------------- | | `google/gemini-3.1-pro-preview` | 2.00 (≤200k)
4.00 (>200k) | 12.00 (≤200k)
18.00 (>200k) | 90% off input | [Gemini 3.1 Pro](https://ai.google.dev/gemini-api/docs/models#gemini-3.1-pro-preview) | | `google/gemini-3.1-flash-lite` | 0.25 | 1.50 | 90% off input | [Gemini 3.1 Flash Lite](https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite) | | `google/gemini-3.1-flash-lite-preview` | 0.25 | 1.50 | 90% off input | [Gemini 3.1 Flash Lite Preview](https://ai.google.dev/gemini-api/docs/models#gemini-3.1-flash-lite-preview) | | `google/gemini-3.5-flash` | 1.50 | 9.00 | 0.15 | [Gemini 3.5 Flash](https://ai.google.dev/gemini-api/docs/models/gemini-3.5-flash) | | `google/gemini-3-flash-preview` | 0.50 | 3.00 | 90% off input | [Gemini 3.0 Flash](https://ai.google.dev/gemini-api/docs/models#gemini-3-flash-preview) | Grok 4.3 and 4.20 variants — reasoning, non-reasoning, and multi-agent. | Model | Input (\$/1M) | Output (\$/1M) | Cache (\$/1M) | Docs | | ----------------------------- | ------------- | -------------- | ------------- | -------------------------------------------------------------- | | `xai/grok-4.3` | 1.25 | 2.50 | 0.20 | [Grok 4.3](https://docs.x.ai/developers/models) | | `xai/grok-4.20-reasoning` | 1.25 | 2.50 | 0.20 | [Grok 4.20 Reasoning](https://docs.x.ai/developers/models) | | `xai/grok-4.20-non-reasoning` | 1.25 | 2.50 | 0.20 | [Grok 4.20 Non Reasoning](https://docs.x.ai/developers/models) | | `xai/grok-4.20-multi-agent` | 1.25 | 2.50 | 0.20 | [Grok 4.20 Multi-Agent](https://docs.x.ai/developers/models) | Nemotron 3 Super — NVIDIA's open-weight reasoning model. | Model | Input (\$/1M) | Output (\$/1M) | Cache (\$/1M) | Docs | | ----------------------------------- | ------------- | -------------- | ------------- | ------------------------------------------------------------------------------ | | `nvidia/nemotron-3-super-120b-a12b` | 0.25 | 2.50 | — | [Nemotron 3 Super 120B](https://research.nvidia.com/labs/nemotron/Nemotron-3/) | Not all third-party models support all features (e.g., reasoning, tools). Check model documentation for specific capabilities. ## Using a Model ```python Python theme={null} from perplexity import Perplexity client = Perplexity() response = client.responses.create( model="openai/gpt-5.5", input="Explain the difference between supervised and unsupervised learning in machine learning.", max_output_tokens=300, ) print(f"Response ID: {response.id}") print(response.output_text) ``` ```typescript Typescript theme={null} import Perplexity from '@perplexity-ai/perplexity_ai'; const client = new Perplexity(); const response = await client.responses.create({ model: "openai/gpt-5.5", input: "Explain the difference between supervised and unsupervised learning in machine learning.", max_output_tokens: 300, }); console.log(`Response ID: ${response.id}`); console.log(response.output_text); ``` ```bash cURL theme={null} curl https://api.perplexity.ai/v1/agent \ -H "Authorization: Bearer $PERPLEXITY_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "openai/gpt-5.5", "input": "Explain the difference between supervised and unsupervised learning in machine learning.", "max_output_tokens": 300 }' | jq ``` **See Your Costs in Real-Time:** Every response includes a `usage` field with exact input tokens, output tokens, and cache read tokens. Calculate your cost instantly using the pricing table above. ```json theme={null} { "usage": { "input_tokens": 150, "output_tokens": 320, "total_tokens": 470 } } ``` ## Model Fallback For high-availability applications, you can specify multiple models in a fallback chain. When one model fails or is unavailable, the API automatically tries the next model in the chain. Learn how to use model fallback chains to ensure high availability and reliability by automatically trying multiple models when one fails. **Example:** ```python theme={null} response = client.responses.create( models=["openai/gpt-5.5", "anthropic/claude-sonnet-4-6", "google/gemini-3-flash-preview"], input="Your question here" ) ``` For detailed examples, pricing information, and best practices, see the [Model Fallback documentation](/docs/agent-api/model-fallback). ## Next Steps Equip your model with web search for source-grounded context. Write prompts that get the most out of the Agent API. Shape responses with structured outputs and JSON schemas. Query market data, filings, and ticker-level information.