Perplexity Orchestrator
Perplexity Launches Orchestrator API for Model Routing and Failover
Perplexity opened its internal model‑routing layer to developers with the Orchestrator API, giving one endpoint for OpenAI, Anthropic, Google, xAI and open‑weight models with automatic same‑model failover and unified billing. Unlike OpenRouter's 5.5% fee, Perplexity charges no platform fee on credits.
One Endpoint for Frontier and Open‑Weight Models
Perplexity has opened the model‑routing layer it already uses in production to outside developers. The Orchestrator API puts OpenAI, Anthropic, Google, xAI and open‑weight models like GLM, Kimi and DeepSeek behind a single endpoint, according to Perplexity. It accepts OpenAI Chat Completions, OpenAI Responses and Anthropic Messages formats, so most teams only need to swap their API key and base URL, per.2
Perplexity says it routes hundreds of millions of requests daily across more than 85 models and over a dozen providers. The company wrote that "through one API, teams can run simple, high‑volume tasks on lower‑cost models and reserve more powerful models for more demanding work."
How Automatic Failover Works
Failover keeps the model fixed and swaps the deployment: a Claude request failing on Anthropic can be served by AWS Bedrock, for instance. It triggers on provider errors, rate limits and timeouts, and requests that fail before producing output are not billed, per.2
1 claims that in one month of production traffic, Claude and GPT requests through Orchestrator had higher success rates than any individual provider, with less daily variation during provider incidents. It's a meaningful claim: a success‑rate bump from 99.0% to 99.99% can save thousands of failed responses for high‑volume teams.
Prompt‑Based Routing in Preview
Orchestrator ships with a preview feature called Routes that lets developers describe in plain language which requests each model should handle. You set a default model, then define up to five routes with a name, description and model. A decision model reads the conversation and route descriptions — never the route names or model IDs — and picks where each request goes, per.2
There's also an Auto Effort preview that dynamically adjusts how much reasoning a model uses on its next turn based on a developer‑defined policy. An agent, for example, can spend more effort recovering from a tool error without switching models.
Pricing: No Platform Fee, Tiered Access
Orchestrator bills at the answering model's published per‑token rate regardless of which provider serves it, with no additional platform fee on credits, per.2 Access is tiered by cumulative credit purchases:
- Tier 0 $0 — Google, xAI and open‑weight models
- Tier 1 $50 — OpenAI and most Anthropic models
- Tier 2 $250 — Claude Opus 5, Opus 5.5 and Fable models
How It Stacks Up Against OpenRouter and Vercel AI Gateway
The comparison that matters for builders is against the incumbents. 2 broke down the differences:
What Builders Should Watch
The no‑fee structure is the headline. OpenRouter charges a 5.5% standard fee on credit purchases, while Perplexity's Orchestrator passes through the model's list price with no markup — which could make it the cheaper default for teams already paying for Claude and GPT usage. The trade‑off is control: Orchestrator does not support bring‑your‑own‑key, and you cannot manually pick a deployment, per.2
1 positions Orchestrator as the first step in its API platform for intelligent model orchestration. For builders tired of managing multiple provider accounts, the pitch is simple: one key, one invoice, and the routing handled for you.
Sources
- 1.Perplexity(perplexity.ai)
- 2.MarkTechPost(marktechpost.com)
Nov 20, 2026
Arizona Opens Criminal Probe Into OpenAI Over Rogue AI Agents in Banks
Arizona Attorney General Kris Mayes announced a criminal investigation into OpenAI and other AI companies over the "apparent infiltration of banks by rogue AI agents." The probe comes as major banks including Morgan Stanley and JPMorgan Chase have already begun blocking AI agents from customer accounts.
Nov 20, 2026
OpenAI Picks Foxconn's FIH Mobile to Build First AI Phone
Foxconn subsidiary FIH Mobile locked in the assembly order for OpenAI's first phone, packing a custom 2nm MediaTek Dimensity 9600 with dual AI processors. Three other Apple suppliers feed the component pipeline. Mass production targets late 2027 or 2028, a direct shot at breaking free from Google and Apple platforms.
Nov 20, 2026
OpenAI Built a Codex-Powered System to Fix ChatGPT Slowdowns
OpenAI revealed an internal system called Performance Factory that uses Codex agents to investigate and fix ChatGPT performance regressions. Over three months it ran more than 1,700 investigations, and roughly 15 shipped fixes were pull requests written entirely by Codex. It is a working preview of how AI coding agents and human engineers can coordinate on production systems.
Related News
Nov 20, 2026
Qualcomm Lays Off Hundreds and Raises $4B in AI Pivot
Qualcomm is laying off roughly 520 employees in California and a reported 1,500 in India while raising $4 billion in debt, part of a pivot away from its shrinking smartphone-chip business toward AI data centers and automotive. CEO Cristiano Amon called the cuts a 'performance-focused reduction,' but nearly half of layoff discussions on Blind cited favouritism.
Nov 19, 2026
Perplexity Brain Creates 80,000 Agent Skills for 20,000 Users From Experience
Perplexity's Brain personalization system has written over 80,000 personal skills for more than 20,000 users, learning workflows from experience. The system reduced tool-call failures by 11% and dissatisfaction threads by 7.6% in production testing.
Nov 15, 2026
Perplexity Teammates AI Agents Arrive on Mac — Behind a $200 Paywall
Perplexity brought its Teammates AI agents to the Mac desktop app this week, but access requires the $200/month Max plan. The always-on agents, launched on the web a week earlier, can perform ongoing research, monitoring, and task execution independently.