Fast Search is now the default search in Hermes Agent for Nous Portal subscribers. It’s built for agentic tasks, with single-search-call latency of 160 ms at p50 and 230 ms at p95. Learn more: pplx.ai/hermes-fast-search

A new Perplexity Search SDK cookbook is live. The recipe fans out focused searches, filters results to official docs, extracts relevant passages, and writes a source-linked brief for your coding agent. Get started: pplx.ai/pplx-srch-sdk

Perplexity has joined the Rust Foundation. We believe in supporting the people who build reliable open-source software. By joining the Rust Foundation, our goal is to improve how people and agents build with Rust.

The Perplexity API MCP server now supports OAuth. Add api.perplexity.ai/mcp to your client, sign in with your Perplexity account, choose your org, and approve. Your agent gets real-time web search, deep research, and advanced reasoning. Get started: pplx.ai/api-get-started

Perplexity Search API is now available in Hermes Agent. Search API gives Hermes access to an index of more than 400 billion URLs. It returns real-time results and ranks snippets by relevance. pplx.ai/hermes

Perplexity API is now available in Stripe Projects. In the Stripe CLI, type "stripe projects add perplexity/api" to get started. Provision a Perplexity API project, API key, and prepaid credits straight from your terminal or coding agent using the Stripe Projects CLI. pplx.ai/pplx-stripe-pro

42 models are now available through the Perplexity Agent API, including the open-weight GLM 5.3 from @Zai_org. Switch models by changing one field in your request. Add fallbacks to keep your agent running if a model becomes unavailable. pplx.ai/agent-api-glm

Connectors are now available in Agent API. Connect your agents to GitHub, Slack, Google Drive, and Datadog without passing a server URL or token on every request. API Group admins can connect each service once for all API Group members. pplx.ai/api-console

Introducing the Perplexity Search SDK. It's an agent-first Python SDK that brings Perplexity's Search as Code approach to your applications. Agents can fan out multiple searches, then filter, dedupe, and rank results in code.

Sonar is moving to the Agent API. The Perplexity Agent API keeps grounded web search, and adds multi-step research, code execution, built-in tools, and access to multiple models through one API. On BrowseComp and WideSearch, Agent API more than doubles the best Sonar score.

.@NVIDIA Nemotron 3.5 Lightning is now available in the Perplexity Agent API 🎉 An open 30B MoE model with 3B active parameters, built for the high-volume execution layer of always-on agents: tool calls, validation, and subagent work. Pair Nemotron Lightning 3.5 with frontier models like Nemotron Ultra for planning, and let Lightning handle the volume. Start building: bit.ly/perplexity_lightning

Access the power of @Kimi_Moonshot K3 in the Perplexity Agent API. Hosted exclusively on U.S.-based servers. Try it out today bit.ly/pplx-kimi-k3

Build with Claude Fable 5 in the Perplexity Agent API. Developers are using Fable 5 for their hardest agent workflows. In our Agent API, Fable comes with search, retrieval, and code execution included. Get started: docs.perplexity.ai/docs/agen…

The Perplexity CLI is now available, giving coding agents the ability to search the web. Copy this to your agent to get set up: "Read: github.com/perplexityai/api-… and install this skill."

@farango77 good question! it's absolutely for non-coding agents to use too. Search as Code is just the architecture we use to run search inside our Agent API. As a user, you don't need to think about all that! Just call our wide research preset + your agent gets wide/deep research results out of the box :)

Narrow search is essentially solved. That's why simple browse/search benchmarks are so saturated. The frontier is wide research: find every qualifying result and back each one with evidence. Today we launched WANDR, a 500-task benchmark built on real knowledge work. It's difficult even for today's most powerful models. Perplexity Agent API's Search as Code architecture excels on WANDR because it allows the model to design the research once and then deterministically execute it at scale without overwhelming the model’s context. Using wide research in your existing agent research workflows? Try out our wide-research preset in Agent API: docs.perplexity.ai/docs/agen… WANDR Research Article: research.perplexity.ai/artic… Github Repo: github.com/perplexityai/wand…

The GPT-5.6 model family is now available on Perplexity's Agent API. Every gain at the frontier compounds through Perplexity's API stack. The GPT-5.6 family sits on the Pareto frontier of our agentic research evals: higher accuracy at lower cost. As such, we've updated our Agent API Presets. xHigh, High, and Medium (our pre-configured presets for long-running research workloads) are now better, faster and cheaper on the GPT-5.6 model family. Check it out today: docs.perplexity.ai/docs/agen…

Perplexity's Agent API presets have been refreshed: faster, smarter, cheaper while maintaining full computer and usage transparency. Most agentic API endpoints charge a fixed per-request rate in the name of simplicity. This masks what's actually happening under the hood. Providers can use as little computer as possible on your request and you pay the same (regardless of what actually ran). On Agent API, it's simple and transparent: > Pick preset mode: fast / low / medium / high depending on task complexity > Pay only for the inference and tools you actually use (direct model provider pricing) > No manual tuning + no model selection required Try it for yourself: docs.perplexity.ai/docs/agen…

@Zai_org's flagship model, GLM-5.2, is now available in Perplexity's Agent API. GLM-5.2 is one of the strongest open-source models for long-horizon coding and agentic workflows. It shines in Agent API, making particularly effective use of our Search as Code architecture. Combine frontier reasoning with real-time programmatic search with just one API call. OpenAI-compatible interface and first-party pricing with no markup. Get started: docs.perplexity.ai/docs/agen…

Search as Code is now available in the Perplexity Agent API. Search as Code matches or beats every competing system across all five leading benchmarks and sets a new cost-performance frontier.

Gemini 3.5 Flash is now available in Perplexity's Agent API.

Grok 4.3 and Gemini 3.1 Flash Lite are now available in the Perplexity Agent API.

Claude Opus 4.7 and GPT-5.5 are now available in the Perplexity Agent API.

Perplexity API credits are now available on AWS Marketplace. Enterprise customers can purchase API credits directly through their AWS account. Credits are applied to your Perplexity API balance and work across all APIs.

You can now use the Perplexity Search API to add real-time web search to any model from any provider via Vercel AI Gateway. No additional API keys required.

New in the Perplexity Search API — Developers can now filter search results by specific time periods or recency. Perfect for tracking the latest updates or exploring historical context.

Last weekend, we hosted our first Perplexity Hackathon in London 🇬🇧 100+ builders spent 24 hours creating tools, agents, and applications powered by the Perplexity API Platform - competing for £8,000 in prizes. From one-shot research pipeline assistants to feature-rich finance dashboards, the creativity was unreal. Huge thanks to everyone who participated and stayed curious 👀

New in the Perplexity Search API — you can now filter searches by specific domains. Query only trusted sources to get focused, verifiable results.

Last night, we hosted API Day at our SF office for @Techweek_ — what started as a small meetup turned into a packed event with 200+ founders, builders & community leaders. Incredible talks from @AravSrinivas and @denisyarats, live demos, and deep convos about the future of AI-powered Search APIs! Stay curious 👀

You can now upload documents and ask Sonar to analyze them — then instantly cross-reference findings with live web search. Get answers grounded in your data and the latest facts.

Android’s fragmented ecosystem means every manufacturer handles audio differently—leading to inconsistent real-time voice performance. To fix this issue, Perplexity Voice Mode has been using SMPL's VAD and AEC technology to deliver clear conversations across Android devices.

Backed by Science – Best Health Project: Scientific verification for social media videos with fact checks and evidence-based analysis.

Briefo – Best Finance Project: Social platform with news summaries and financial insights from Sonar, for people who want info without the noise.

Walter Wego – Best Deep Research Project: 21st-century intelligence dashboard for deep dives—visualize, map, and synthesize research with real-time data from Sonar.

What Does The World Think? – Most Fun / Creative Project: Global opinion analyzer that visualizes how different countries react to any claim with Sonar.

Therefore — Runner Up: Sonar powered platform to turn health tracking into structured, personal N-of-1 experiments and actionable science.

Vonar AI — First Place: Voice agent that handles calls, support, and real-time answers through Sonar without a human.

Finance with Perplexity Sonar just got a whole lot better. You can now search over SEC filings and get real-time stock data all in a single API request. Market insights, earnings, and disclosures are delivered instantly, ready to use.

Our new Sonar Resources page is live. See what others are building, read articles, and register for upcoming events.

Last month, we had 70+ builders in our SF office for Sonar API Demo Night! We saw 8 incredible demos featuring AI accounting tools, smart real estate assistants, real-time brand monitoring, and more, all built on our Sonar APIs. Didn't get a chance to demo? Join our brand new dev community and share your project and thoughts about our APIs! (link below)

We are actively hiring for the following engineering roles at Perplexity: AI Machine Learning Engineer - Personalization Frontend Engineer - Product Staff Software Engineer - Authentication & Identity Fullstack Engineer - Growth Backend Engineer - Billing AI Inference Engineer

Join us for our official Sonar API Demo Night with @cerebral_valley. This invite-only event will feature Sonar demos from top builders, with opening remarks from @AravSrinivas.

To minimize overhead, we implemented computation and communication overlapping. Our micro-batching approach cuts multi-node communication overhead by up to 40%.

Our implementation leverages three parallelism strategies: • Expert Parallelism for distributing MoE layers • Tensor Parallelism for MLA computation • Data Parallelism for independent inference engines

DeepSeek-V3/R1 contains 671B total parameters but activates only 37B per token. Testing shows EP128 configurations deliver up to 5x higher throughput at equivalent output speeds compared to single-node deployments. Higher EP values assign fewer experts per GPU, reducing memory bandwidth pressure.

Our recent article demonstrates that, contrary to conventional systems, MoE models like DeepSeek-V3/R1 can simultaneously achieve higher throughput and lower latency when utilizing more GPUs in multi-node deployments across most scenarios.

Our first Perplexity hackathon is now live on @devpost! Build a web-enabled app or project using Sonar for a chance to win a share of over $35k in prizes. The deadline is May 28, 2025, at 12 PM PT.