Perplexity Updates

Everything Perplexity has shipped, in one place.

Last Week

Perplexityapi

Link·Aug 20, 2026

41 Models in the Perplexity Agent API

Perplexity's Agent API now reaches 41 frontier models from nine providers through a single endpoint, with search and code execution built in.

Best for: building an agent that routes different jobs to different models without wiring up each provider
When to use it: you want model choice but not nine sets of credentials and nine SDKs
Who benefits: developers building multi-model agent workflows

Perplexityapi

Link·Aug 18, 2026

Perplexity API MCP Server as an Agent Plugin

Perplexity's API connector now installs as an Agent Plugin in coding tools like Codex, Cursor, and VS Code.

Best for: Giving a coding assistant live web search without wiring the connection yourself
When to use it: Your assistant keeps answering from stale training data and needs current sources
Who benefits: Developers running agents inside Codex, Cursor, or VS Code

Perplexityfeature

Link·Aug 18, 2026

Perplexity Computer for Email

You can start a Computer task by emailing it, forwarding a message to it, or adding it to a thread you are already on.

Best for: Kicking off work without leaving your inbox or opening the app
When to use it: The task already lives in an email thread and re-explaining it would take longer than forwarding it
Who benefits: People who work out of their inbox and want the thread itself to be the brief

Perplexitymodel

Link·Aug 18, 2026

DeepSeek V4 Pro in Perplexity Computer

Perplexity added DeepSeek V4 Pro, hosted in the United States, as a model option inside Computer.

Best for: Running Computer tasks at a lower cost each
When to use it: Cost per task matters more to you than topping a benchmark
Who benefits: Teams running Computer at volume and watching spend

August 2026

Perplexityapi

Link·Aug 14, 2026

Perplexity Search SDK

A Python toolkit that lets an AI agent run and refine its own web searches in code instead of making one search call and taking what comes back.

Best for: building an agent that researches a topic across many sources before it answers
When to use it: a single search call is too blunt and you want the agent to run several searches, then sort and trim the results itself
Who benefits: developers building research or monitoring agents on Perplexity

Perplexityfeature

Link·Aug 13, 2026

Search as Code Optimizations

Small update. Perplexity tuned Search as Code, its research engine from June, to run better and cost less per task.

Best for: existing Search as Code workloads, with nothing to change on your side
When to use it: nothing to switch on, the change rolls out this week
Who benefits: developers already running wide or deep research through Search as Code

Perplexityapi

Link·Aug 13, 2026

Sonar Moving to the Agent API

Perplexity is folding its Sonar search API into the Agent API, which keeps grounded web search and adds tools and model choice.

Best for: teams already calling Sonar for grounded web search inside their own products
When to use it: you are starting a new Perplexity API build, or your Sonar calls need more than one search-and-answer step
Who benefits: developers running Perplexity search in production

Perplexitymodel

Link·Aug 13, 2026

Grok 4.6 in Perplexity and Perplexity Computer

xAI's Grok 4.6 is now selectable in Perplexity and in Perplexity Computer.

Best for: running the same Perplexity work on a cheaper model without giving up much quality
When to use it: cost per task matters more to you than squeezing out the last bit of quality
Who benefits: Perplexity and Perplexity Computer users who pick their own model

Perplexitymodel

Link·Aug 12, 2026

Nemotron 3.5 Lightning in the Perplexity Agent API

NVIDIA's open-weight Nemotron 3.5 Lightning is now available in Perplexity's Agent API, priced for the high-volume, repetitive parts of an agent's work.

Best for: The cheap, repetitive steps inside an agent: calling tools, checking results, and handing work off to subagents.
When to use it: An agent is making the same kind of call thousands of times and a frontier model's price stops making sense.
Who benefits: Developers running always-on agents where the constraint is cost per call, not reasoning depth.

Perplexityapi

Link·Aug 10, 2026

Kimi K3 in the Perplexity Agent API

Moonshot AI's Kimi K3 model is now callable through Perplexity's Agent API, hosted on U.S. servers.

Best for: Building agents on a model outside the big three labs without routing data to overseas servers.
When to use it: You want Kimi K3 inside an agent workflow and the hosting location matters to your team.
Who benefits: Developers on the Perplexity Agent API comparing model cost against capability.