API Changelog

The Decisions API returns probabilities instead of generated text via POST /v1/decisions with pplx-decider-v1-27b for yes-or-no, choice, or scored answers. The Agent API now accepts PDF, DOC, DOCX, TXT, and RTF documents as input_file content, adds an image_search tool at $2.50 per 1,000 invocations, and reports token usage and cost for pplx_sdk.llm.extract calls inside the sandbox tool. The low, medium, and high presets now use Fast Search, cutting web_search costs from $2.50 to $1.00 per 1,000 invocations.

Agent API now supports openai/gpt-6.1-sol, anthropic/claude-sonnet-5-5, google/gemini-3.8-flash with promotional pricing through Dec 31, and perplexity/glm-5.3-flash at $0.15 per million uncached-input tokens, while seven OpenAI GPT-5-family models retire on October 24, 2026 at 00:00 UTC. Also added a lower-latency Fast Search path via search_type: "fast" at $1.00 per 1,000 requests, custom MCP connectors that store credentials once per project, and OAuth sign-in for the remote Perplexity MCP Server.

Agent API and Router API now support perplexity/glm-5.3, xai/grok-4.6, and perplexity/nemotron-3-ultra-550b-a55b. Agent API presets use stable prompt cache keys to reuse shared prompt prefixes and cut costs by about 5%, while Gemini 3.7 Flash pricing rose on August 27, 2026 to $0.75 per million input tokens, $0.075 per million cached-input tokens, and $3.75 per million output/reasoning tokens.

Agent API now supports the GPT-5.6 family (sol, terra, luna) and Grok 4.5, all with first-party token pricing.

Agent API now supports Claude Sonnet 5, GLM 5.2, Kimi K2.7 Code, and Nemotron 3 Super, all with direct first-party token pricing.

The Agent API added support for Claude Opus 4.8, Gemini 3.5 Flash, Gemini 3.1 Flash Lite, and the Grok 4.3 and 4.20 family, all with direct first-party token pricing.

The finance_search tool is now available in the Agent API. Pull structured financial and market data — quotes, financials, earnings, analyst estimates, segment KPIs, ETF constituents, and more — for public companies and instruments. The model decides which fields to fetch…

Agent API: New Third-Party Models — Claude Opus 4.7, GPT-5.5, and Grok 4.20 Reasoning now supported.

Agent API: New Third-Party Models — GPT-5.4, NVIDIA Nemotron, Claude Sonnet 4.6, and Gemini 3.1 Pro Preview now supported.

The Agent API is now generally available. Build autonomous agents with production-ready guidance on model behavior, output controls, and OpenAI-compatibility patterns.

The Embeddings API is now generally available with guides for standard and contextualized embeddings, plus best practices for semantic search and retrieval workflows.

Model Deprecation: sonar-reasoning Removed — As of December 15, 2025, sonar-reasoning has been deprecated. Migrate to sonar-reasoning-pro.

The Perplexity MCP Server now supports one-click installation for Cursor, VS Code, Claude Desktop, and Claude Code. Provides four tools: perplexity_search, perplexity_ask, perplexity_research, and perplexity_reason.

Pro Search is now generally available for Sonar Pro. Enhances queries with automated tool usage and multi-step reasoning through intelligent tool orchestration.

Official Perplexity SDKs for Python and TypeScript are now available.

We're thrilled to announce the official Perplexity SDKs for Python and Typescript! These SDKs provide convenient, type-safe access to all Perplexity APIs with both synchronous and asynchronous clients.

• language_preference: Specify preferred languages for search results (available for sonar and sonar-pro)
• search_domain_filter: Filter results to specific domains
• Date/time filters: Enhanced control over result freshness with publication and update filters

New interactive playground to test Search API queries and parameters in real time — no API key required to get started. Experiment with filtering options, see response structures, and refine queries before implementing in code.

Standalone Search API providing direct access to search results without LLM processing. Returns raw, ranked search results from Perplexity's continuously refreshed index.

Upload and analyze documents in multiple formats using Sonar models. Supports PDF, DOC, DOCX, TXT, and RTF files.