惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

G
Google Developers Blog
S
Schneier on Security
The Hacker News
The Hacker News
P
Proofpoint News Feed
Spread Privacy
Spread Privacy
L
LINUX DO - 热门话题
L
Lohrmann on Cybersecurity
I
Intezer
cs.AI updates on arXiv.org
cs.AI updates on arXiv.org
www.infosecurity-magazine.com
www.infosecurity-magazine.com
Schneier on Security
Schneier on Security
Security Latest
Security Latest
AWS News Blog
AWS News Blog
B
Blog RSS Feed
Microsoft Security Blog
Microsoft Security Blog
有赞技术团队
有赞技术团队
博客园 - 叶小钗
The Last Watchdog
The Last Watchdog
O
OpenAI News
月光博客
月光博客
Hacker News: Ask HN
Hacker News: Ask HN
阮一峰的网络日志
阮一峰的网络日志
S
Security @ Cisco Blogs
Google Online Security Blog
Google Online Security Blog
云风的 BLOG
云风的 BLOG
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
Latest news
Latest news
P
Palo Alto Networks Blog
Last Week in AI
Last Week in AI
M
MIT News - Artificial intelligence
Google DeepMind News
Google DeepMind News
P
Proofpoint News Feed
C
CERT Recently Published Vulnerability Notes
Apple Machine Learning Research
Apple Machine Learning Research
U
Unit 42
PCI Perspectives
PCI Perspectives
博客园 - 聂微东
SecWiki News
SecWiki News
宝玉的分享
宝玉的分享
Forbes - Security
Forbes - Security
H
Heimdal Security Blog
cs.CL updates on arXiv.org
cs.CL updates on arXiv.org
Hugging Face - Blog
Hugging Face - Blog
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
T
Troy Hunt's Blog
博客园 - 三生石上(FineUI控件)
Application and Cybersecurity Blog
Application and Cybersecurity Blog
罗磊的独立博客
WordPress大学
WordPress大学
D
Darknet – Hacking Tools, Hacker News & Cyber Security

OpenRouter Blog

Choosing the Optimal Image Input Detail Level in LLMs — OpenRouter Blog DeepSeek V4 Is Earning Agentic Token Share — OpenRouter Blog The Open Weight Models that Matter: June 2026 — OpenRouter Blog The OpenRouter MCP Server — OpenRouter Blog Introducing the Unified Image API — OpenRouter Blog The AI Governance Checklist That Maps to Your Stack — OpenRouter Blog Enforce AI Data Residency at the Routing Layer — OpenRouter Blog OpenRouter vs LiteLLM: Managed vs Self-Hosted Gateway — OpenRouter Blog Connect OpenClaw to OpenRouter: One Key, Failover, Free Models — OpenRouter Blog Connect SillyTavern to OpenRouter: Setup, Models, Fixes — OpenRouter Blog A Robot is Sprinting Towards You: Do You Want it Running on Claude or Grok? Kilo Code + OpenRouter: Setup, Routing, and Free Models — OpenRouter Blog Codex CLI with OpenRouter: config.toml Setup and Models — OpenRouter Blog Claude Code with OpenRouter: Setup, Models, and Costs — OpenRouter Blog How to Use OpenRouter With Any Coding Agent or AI Tool — OpenRouter Blog Subagent: Let Your Model Delegate the Busywork — OpenRouter Blog Free LLM API in 2026: 13 Options Ranked and Compared — OpenRouter Blog How to Enforce Agentic AI Governance at the API Layer — OpenRouter Blog Keep Your Agent Running When Models Disappear — OpenRouter Blog Hermes Agent + OpenRouter: Setup, Model Choice & Routing Config — OpenRouter Blog Lowest-Cost LLM Inference: The Complete OpenRouter Guide — OpenRouter Blog How OpenRouter Model Routing Works: Providers, Fallbacks & Auto Router — OpenRouter Blog OpenRouter Failover: Provider Failover vs Model Fallbacks Explained — OpenRouter Blog Surpassing Frontier Performance with Fusion — OpenRouter Blog Dinner is Served — OpenRouter Blog LLM Gateway: What It Is and How to Choose One — OpenRouter Blog Advisor: Give Any Model a Lifeline to a Smarter One — OpenRouter Blog Gemini 2.5 Flash API - Pricing, Quickstart & Provider Comparison — OpenRouter Blog EU AI Act & Colorado ADMT Compliance: Human Oversight for AI Agents — OpenRouter Blog May Release Spotlight — OpenRouter Blog Guardrails: Protect your Agents, Data, and Costs — OpenRouter Blog OpenRouter Raises $113M Series B — OpenRouter Blog Human-in-the-Loop Tools for the Agent SDK — OpenRouter Blog Consistent Web Search and Fetch Across Every Model — OpenRouter Blog GPT-5.5 Price Increase: What It Actually Costs — OpenRouter Blog New Audio APIs for Speech and Transcription — OpenRouter Blog Response Caching: Zero Cost for Identical Requests — OpenRouter Blog April Release Spotlight — OpenRouter Blog Create OpenRouter Accounts via CLI with Stripe Projects — OpenRouter Blog Opus 4.7 Agent SDK: Building Multi-turn Agent Workflows on OpenRouter — OpenRouter Blog Build Your Own Harness with the Agent SDK — OpenRouter Blog Introducing Workspaces — OpenRouter Blog Announcing Video Generation — OpenRouter Blog Auto Exacto: Adaptive Quality Routing, On by Default — OpenRouter Blog February Release Spotlight — OpenRouter Blog OpenRouter Outages on February 17 and 19, 2026 — OpenRouter Blog January Release Spotlight — OpenRouter Blog Distillable Models and Synthetic Data Pipelines with NeMo Data Designer — OpenRouter Blog December Release Spotlight — OpenRouter Blog Response Healing: Reduce JSON Defects by 80%+ — OpenRouter Blog The 2025 State of AI Report — OpenRouter Blog Is Implicit Caching Prompt Retention? — OpenRouter Blog Provider Variance: Introducing Exacto — OpenRouter Blog 1 million free BYOK requests per month — OpenRouter Blog The First-Ever Image Model Is Up on OpenRouter — OpenRouter Blog GPT-5 is now live — OpenRouter Blog Audio Inputs and PDF URLs for Apps — OpenRouter Blog Presets: How To Seamlessly Transfer Model Configurations Across Apps — OpenRouter Blog New Privacy-Focused Provider Drop: Venice — OpenRouter Blog Use OpenRouter Models in Cursor: Try it with Moonshot AI Updates to Our Free Tier: Sustaining Accessible AI for Everyone — OpenRouter Blog New Stealth Model: "Cypher Alpha" — OpenRouter Blog Introducing Presets: Manage LLM Configs from Your Dashboard! — OpenRouter Blog Dev & BYOK Updates: Uptime API + Smarter Key Management — OpenRouter Blog Simplifying Our Platform Fee — OpenRouter Blog GIF Prompts, Omni Search, Tool Caching, and BYOK Flags — OpenRouter Blog New Features: Reasoning Streams, Crypto Invoices, End-User IDs & More — OpenRouter Blog Passkeys, DevEx Upgrades, and a New Guide for TypeScript Agents — OpenRouter Blog New Provider Drop: Cerebras Is Here — OpenRouter Blog Better Insights, Faster Metrics, and New Developer Power Tools — OpenRouter Blog Privacy Clarity, New Providers, OAuth Upgrade, and Gemini Gets Parallel Tools — OpenRouter Blog Universal PDF Support — OpenRouter Blog Smarter Charts, Inline SVGs, and Live Usage Accounting — OpenRouter Blog Quasar Alpha and Optimus Alpha Reveal — OpenRouter Blog "Stealth" model: Optimus Alpha — OpenRouter Blog “Stealth” model: Quasar Alpha — OpenRouter Blog Never Pay for Empty AI Responses Again — OpenRouter Blog Deep Research & Many New Models — OpenRouter Blog Introducing Nitro and Floor Price Shortcuts — OpenRouter Blog Introducing Cloudflare as a new provider — OpenRouter Blog Reasoning Tokens for Thinking Models — OpenRouter Blog Introducing Web Search via the API — OpenRouter Blog Standardized finish reasons — OpenRouter Blog Happy New Year! Introducing a new Auto Router — OpenRouter Blog Holiday launches: Web Search & Price Cuts — OpenRouter Blog Bring Your Own API Keys — OpenRouter Blog Crypto Payments API — OpenRouter Blog Structured Outputs & Free Gemini Flash 2.0 — OpenRouter Blog Price Drops and Llama 3.3 70b — OpenRouter Blog Author Pages & Amazon Nova — OpenRouter Blog
OpenRouter vs Portkey: Routing Network vs Control Plane — OpenRouter Blog
OpenRouter · 2026-06-20 · via OpenRouter Blog

If you’re comparing OpenRouter and Portkey, the real question is whether you want something that routes requests for you, or something that governs the requests you already make.

OpenRouter is a managed routing network. You buy credits, call one OpenAI-compatible endpoint, and it routes across 70+ providers with automatic failover. You never hold a provider key. Portkey is an AI control plane, now part of Palo Alto Networks after a 2026 acquisition. It sits in front of the provider keys you already have and adds governance, prompt management, guardrails, and observability.

Pick OpenRouter when you want model access and failover with no infrastructure. Pick Portkey when you need a governance layer over the provider accounts you already run.

Route across providers, or govern your own keys

OpenRouter owns the path to the model. Your app calls https://openrouter.ai/api/v1, OpenRouter picks a provider, and forwards the request on credits you’ve already bought. There are no per-provider accounts to manage, and a request that fails on one provider falls over to another automatically.

Portkey owns the layer in front of your keys. You bring your own provider credentials (OpenAI, Anthropic, Azure, and the rest), store them as virtual keys, and Portkey routes, caches, and logs every call. Its open-source gateway has 12K+ GitHub stars and installs in 3 lines of code, and you can self-host it or run the managed version.

With OpenRouter you offload provider relationships entirely; with Portkey you keep them and wrap them in policy.

Count models, but count the right thing

Portkey advertises access to 1,600+ LLMs through its unified API. OpenRouter lists 300+ models across 70+ providers, including 20+ free models for evaluation. The raw numbers favor Portkey, but they measure different things.

Portkey’s count is breadth of models you can address once you’ve supplied the keys. OpenRouter’s count is models it actively routes to, with credits, load balancing, and cross-provider failover handled for you. Its Auto Router, powered by NotDiamond, can even pick a model per prompt, and provider routing lets you filter by price, throughput, latency, data policy, and quantization.

Count whichever number you like. The decision still rests on whether you want a network that routes for you or a unified front end over the provider accounts you run.

Match observability and governance to your team

Portkey centers on governance. It ships in-product logs, traces, and dashboards with tiered retention (3 days on the free Developer plan, 30 days on Production, custom on Enterprise), plus prompt templates with a playground, role-based access control, budget and rate limits, and guardrails that include PII redaction. For a platform team standardizing many internal apps, that control plane is the point.

OpenRouter takes a lighter approach to observability. Rather than store your logs long-term, Broadcast streams traces to the platforms you already run, including Datadog, Langfuse, LangSmith, Braintrust, OpenTelemetry, and S3. Workspaces add per-team organization and budgets for teams that want structure across many internal apps.

On compliance, both hold SOC 2 Type 2 and support GDPR. OpenRouter adds Zero Data Retention and EU in-region routing for enterprise accounts, with its report at trust.openrouter.ai. Portkey adds HIPAA and custom BAAs on Enterprise, along with VPC hosting and data isolation. Teams that want data residency without running infrastructure point to OpenRouter.

Pick a pricing model that matches your volume

The two price along different axes. OpenRouter passes provider pricing through at 0% markup, then charges a 5.5% platform fee on credit purchases with a $0.80 minimum. Bring Your Own Key drops that to 5%, and the first 1 million requests each month are waived. You pay for usage, nothing else.

Portkey prices on logs and features. Its Developer tier is free with 10K recorded logs a month, though Portkey marks it as unsuitable for production. Production is $49/month for 100K logs (plus $9 per additional 100K up to 3 million) and adds RBAC and guardrails. Enterprise is custom, with HIPAA, SSO, private deployment, and longer retention. The open-source gateway is free to self-host if you’d rather run it yourself.

Net cost depends on where your weight sits. Heavy model spend with provider keys already in hand often favors Portkey’s self-hosted gateway; usage you’d rather not meter or operate favors OpenRouter’s pay-as-you-go.

Use both, or start where the pain is

These aren’t mutually exclusive. You can point Portkey at OpenRouter as an upstream provider, putting Portkey’s governance and prompt management in front of OpenRouter’s routing network. OpenRouter also lists Portkey as an upcoming broadcast destination, so traces will flow the other direction too.

If you’re weighing more options, our LLM gateway comparison covers LiteLLM, Helicone, Cloudflare AI Gateway, and others.

Frequently asked questions

Is Portkey the same kind of tool as OpenRouter?

They overlap on a unified API but solve different problems. OpenRouter is a managed routing network: you buy credits and it routes across 70+ providers with automatic failover, no provider keys required. Portkey is a control plane that sits in front of the provider keys you already hold, adding governance, prompt management, and in-product observability.

Did Palo Alto Networks acquire Portkey?

Yes. Palo Alto Networks completed its acquisition of Portkey in 2026, folding it into the company’s AI security portfolio. Portkey continues to operate its gateway, including the open-source version.

How do OpenRouter and Portkey compare on compliance?

Both hold SOC 2 Type 2 and support GDPR. OpenRouter adds Zero Data Retention and EU in-region routing for enterprise accounts. Portkey offers HIPAA and custom BAAs on its Enterprise plan. For specific certification requirements, check each platform’s trust center.

Which one is cheaper?

It depends on volume. OpenRouter charges a 5.5% platform fee on credit purchases (5% with Bring Your Own Key, first 1 million requests each month waived) and 0% markup on provider prices. Portkey is free to self-host or on its Developer tier, then $49/month for Production. High-volume teams that already hold provider keys often run Portkey’s open-source gateway; teams that want managed routing without operating anything pick OpenRouter.

Can I use OpenRouter and Portkey together?

Yes. You can point Portkey at OpenRouter as one of its upstream providers, putting Portkey’s governance and observability in front of OpenRouter’s routing network. OpenRouter also lists Portkey as an upcoming broadcast destination for sending traces.