Perplexity 推出 Orchestrator API,支持故障转移与模型路由
Perplexity 推出了 Perplexity Orchestrator API,这是一个面向前沿模型和开放权重模型的统一端点。它根据可用性、容量和延迟选择服务提供商,并通过 1 个账户统一计费。开发者需要做的集成更少,只需处理 1 张账单。
Perplexity Orchestrator API 是一个托管式推理网关,在 1 个基础 URL 背后提供多种模型的服务。它涵盖来自 OpenAI、Anthropic、Google 和 xAI 的前沿模型,也提供 GLM、Kimi 和 DeepSeek 等开放权重模型,这些模型由 Perplexity 或推理合作伙伴托管。
它接受 OpenAI Chat Completions、OpenAI Responses 和 Anthropic Messages 请求。大多数团队只需更改 API key 和基础 URL。OpenAI 的基础 URL 是 https://api.perplexity.ai/orchestrator/v1。Anthropic SDK 用户则指向 https://api.perplexity.ai/orchestrator,不带 /v1。
Perplexity 在内部运行着同一层,覆盖 85+ 个模型和十多家提供商。
故障转移会保持模型不变,只更换部署。一个在 Anthropic 上失败的 Claude 请求,可以改由 AWS Bedrock 提供服务。可靠性文档规定了这些规则:
你无法自行选择部署。
Perplexity 统计了 2026 年 10 月 4 日至 11 月 2 日的生产流量。通过 Perplexity 发出的 Claude 和 GPT 请求,成功率高于所测得的任何单一提供商。在提供商发生事故期间,每日成功率的波动也更小。该文章没有给出具体的成功率百分比。
速度也因提供商而异。对于 Claude Opus 5.5,Anthropic 在 30 天中有 15 天最快,AWS Bedrock 为 15 天。Azure 对 GPT-5.6 Sol 的输出速度中位数比 OpenAI 高 27%。速度感知路由仍在推进中。
路由功能处于预览阶段。你将 model 设为 perplexity/auto,并传入一个 routing 对象。该对象包含一个 default_model 和 1 到 5 条路由,每条路由都有 name、description 和 model。
决策模型会读取对话和路由描述。它从不会看到路由名称或模型 ID。如果没有匹配项,或者决策模型失败或超时,则由默认模型作答。媒体输入也是如此。
响应头会报告结果:x-perplexity-autoroute-outcome、-route 和 -confidence。默认策略将流量发送到 GPT-6.1 Sol,简单请求则流向 GPT-6 Luna。Perplexity 还列出了内置的 Lower cost、Balanced 和 Higher quality 配置文件,以及一个 Auto effort 预览。
Perplexity 在 Terminal-Bench 4.1 的 63 项任务子集上测试了该功能,运行于 Claude Code 中。Anthropic 记录称,Claude Fable 5.2 的防护机制可能会阻止良性的安全和生命科学工作。
这些路由覆盖了安全、生物学和 after_refusal 请求。这是 +3 项任务,且每项任务的估计成本更高。Perplexity 称其为调整后的比较;结果因工作负载而异。
每个请求按应答模型公布的每 Token 费率计费。故障转移到另一个提供商不会改变费率。Perplexity 表示,在额度之外没有额外的平台费用。
例如,perplexity/glm-5.3 列出每 1M Token 输入 $1.4、输出 $4.4 和缓存读取 $0.26。用量从预付的 Perplexity API 额度中扣除,这些额度也可通过 AWS Marketplace 购买。
访问权限按累计额度购买量分级:
请查看发布文章、技术博客、快速上手指南、模型目录和路由文档。你可以通过我们的 GPULeaderboard.com Chrome 扩展程序在这里查看部署选项和规格。
需要与我们合作推广你的 GitHub 仓库、Hugging Face 页面、产品发布或网络研讨会等?欢迎联系我们。
文章《Perplexity Launches Orchestrator API for Failover and Model Routing》首发于 MarkTechPost。
译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。
查看英文原文(Perplexity Launches Orchestrator API for Failover and Model Routing)
Perplexity has launched the Perplexity Orchestrator API, a single endpoint for frontier and open-weight models. It picks a serving provider by availability, capacity and latency, and bills everything through 1 account. Developers get fewer integrations and 1 invoice.
What is the Perplexity Orchestrator API?
The Perplexity Orchestrator API is a hosted inference gateway that serves many models behind 1 base URL. It covers frontier models from OpenAI, Anthropic, Google and xAI. It also serves open-weight models such as GLM, Kimi and DeepSeek, hosted by Perplexity or inference partners.
It accepts OpenAI Chat Completions, OpenAI Responses and Anthropic Messages requests. Most teams only change the API key and base URL. The OpenAI base URL is https://api.perplexity.ai/orchestrator/v1. Anthropic SDK users point at https://api.perplexity.ai/orchestrator, without /v1.
Perplexity runs the same layer internally, across 85+ models and more than a dozen providers.
How does provider failover work?
Failover keeps the model fixed and swaps the deployment. A Claude request that fails on Anthropic can be served by AWS Bedrock instead. The reliability docs set these rules:
- Triggers are provider errors, rate limits and timeouts.
- Each attempt has a time-to-first-token limit and a total-duration limit. The values are not published.
- Invalid or over-context requests return HTTP 400 and are not retried.
- If every deployment fails, the API returns HTTP 429 with
Retry-After. - Requests that fail before producing output are not billed.
You cannot pick a deployment yourself.
What do the reliability and speed numbers show?
Perplexity measured production traffic from October 4 to November 2, 2026. Claude and GPT requests through Perplexity succeeded more often than requests to any single provider measured. Daily success rates also varied less during provider incidents. The post gives no percentage success rates.
Speed varied by provider too. For Claude Opus 5.5, Anthropic was fastest on 15 of 30 days, AWS Bedrock on 15. Azure’s median output speed for GPT-5.6 Sol was 27% higher than OpenAI’s. Speed-aware routing is still in progress.
How does automatic routing with perplexity/auto work?
Routing is in preview. You set model to perplexity/auto and pass a routing object. It holds a default_model and 1 to 5 routes, each with a name, description and model.
A decision model reads the conversation and route descriptions. It never sees route names or model IDs. If nothing matches, or the decision model fails or times out, the default model answers. So do media inputs.
Response headers report the result: x-perplexity-autoroute-outcome, -route and -confidence. The default policy sends traffic to GPT-6.1 Sol, with simple requests going to GPT-6 Luna. Perplexity also lists built-in Lower cost, Balanced and Higher quality profiles, plus an Auto effort preview.
Can routing work around refusals?
Perplexity tested this on a 63-task subset of Terminal-Bench 4.1, run in Claude Code. Anthropic documents that Claude Fable 5.2 safeguards can block benign security and life sciences work.
- Claude Fable 5.2 alone: 38/63 tasks.
- Fable plus routes to GPT-6.1 Sol: 41/63.
- Fable plus routes to GPT-6 Astra: 41/63.
The routes covered security, biology and after_refusal requests. That is +3 tasks, at a higher estimated cost per task. Perplexity calls it an adjusted comparison; results vary by workload.
How is Orchestrator priced?
Each request is billed at the answering model’s published per-token rate. Failover to another provider does not change the rate. Perplexity says there is no additional platform fee on top of credits.
For example, perplexity/glm-5.3 lists $1.4 input, $4.4 output and $0.26 cache read per 1M tokens. Usage draws from prepaid Perplexity API credits, also purchasable through AWS Marketplace.
Access is tiered by cumulative credit purchases:
- Tier 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 the Fable models.
How does it compare with OpenRouter and Vercel AI Gateway?
| Feature | Perplexity Orchestrator | OpenRouter | Vercel AI Gateway |
|---|---|---|---|
| Model catalog | Frontier (OpenAI, Anthropic, Google, xAI) plus open-weight GLM, Kimi, DeepSeek; count not disclosed (blog) | 400+ models, 70+ providers (source) | 400+ models (source) |
| API formats | Chat Completions, Responses, Anthropic Messages (docs) | Chat Completions, Responses, Anthropic Messages (docs) | Chat Completions, Responses, Anthropic Messages (docs) |
| Platform fee | No additional platform fee on credits (blog) | 5.5% on credit purchases (Standard), 8% (Business) (pricing) | No platform fee, no token markup (pricing) |
| Token pricing | Answering model’s published rate (FAQ) | Provider price passed through (FAQ) | Provider list price (docs) |
| Same-model provider failover | Yes, automatic (docs) | Yes, automatic fallback (source) | Yes, automatic (changelog) |
| Prompt-based auto router | perplexity/auto, preview, 1 to 5 routes (docs) | openrouter/auto, no router fee (docs) | Not disclosed |
| Bring your own key | Not supported (FAQ) | Supported, 5% fee above allowance (FAQ) | Supported, no markup (docs) |
Key Takeaways
- 1 endpoint for OpenAI, Anthropic, Google, xAI and open-weight models.
- Same-model failover across providers, billed at the same rate.
perplexity/autoroutes by plain-language descriptions, in preview.- Refusal routing: 38/63 to 41/63 on Terminal-Bench 4.1, self-reported.
Check out the launch post, technical blog, quickstart, model catalog and routing docs. You can check the deployment options and specs here through our GPULeaderboard.com chrome extension.
Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us
The post Perplexity Launches Orchestrator API for Failover and Model Routing appeared first on MarkTechPost.
正文保留英文原文,标题与摘要由机器翻译。
原文发布于 MarkTechPost(Perplexity Launches Orchestrator API for Failover and Model Routing),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。