If you're working on AI agents, you know the challenges—reliability, cost, and orchestration. Portkey helps companies bridge these gaps, making AI agents enterprise-ready.

Meet Ayush Garg (Co-Founder & CTO) at the summit

📅 19-22 Feb | NYC

Jan is a simple open-source app that works as ChatGPT alternative. With Portkey, you get access to 1600+ models, plus built-in observability and governance—right from the start.

Try it out today: Link to docs 

Run your entire LLM stack where your enterprise apps live - Same infrastructure, same compliance standards (SOC2, HIPAA, GDPR), zero friction.

Enterprise AI just got smoother. Deploy through your Azure setup today.

 Link 

Anything LLM (36k+ ⭐) is one of the most loved open-source apps—chat with any LLM, use RAG, and build AI Agents, all in one place.

With Portkey, it’s now simpler, more secure, and scalable.

✨ One integration, 1600+ models – From OpenAI and Claude to your own Ollama deployments.

🔒 Granular security & access control – Set budget & rate limits, mask keys, track usage, and enforce RBAC.

👥 Enterprise-grade AI operations – Manage teams, workspaces, metadata logging, and audit logs with ease.

          Link to Docs 

You can now use Qdrant vector database with Portkey. With this integration, you get:

• Unified API - Call Qdrant alongside your LLMs with Portkey's unified API

• Virtual Keys - Manage your API keys effortlessly

Streamline your AI workflows today!

 Link to docs 

🚀 Portkey now integrates with Zed natively!

Zed is a next-gen code editor with built-in AI features. Portkey's integration with Zed gets you:

• Access 1600+ LLMs

• Real-time observability & logging

• Enterprise governance

• PII detection & redaction

 Link to docs 

LLM costs can escalate quickly. Currently, there is no way to track granular insights on the cost of your AI app. With Portkey’s metadata-based tracking, you get:

✅ Per-user, team, or feature-level cost insights

✅ Dashboards for real-time tracking

✅ Portkey API to fetch and analyze spend programmatically

 Link to cookbook 

Azure is sunsetting all GPT-35 Turbo model versions 0313 and 0613 on 13th February 2025. This means a migration is required! The default model will be set to version 0125 after the retirement.

Quick action items:

✅ Audit your Azzure API calls

✅ Plan migration to Gpt-4 series

✅ Finish all tests before February

OpenAI's latest & greatest o3-mini model is now available on Portkey on both OpenAI & Azure OpenAI!

✅ Backward compatible with system role. (On Portkey you can continue using system role and we convert it to developer for the reasoning models)

✅ Backward compatible with max_tokens param. (we convert it to max_completion_tokens at the Gateway level)

✅ Route to both OpenAI & Azure OpenAI

✅ Log your requests and get metrics on cost, latency, and more

✅ Enforce model behavior with real-time guardrails

 Try it out now! → 

🚀 Want to use Deepseek R1 in the US? Now you can!

We've teamed up with Fireworks AI in our latest cookbook to make R1 accessible via Portkey’s unified API.

Get the best of both worlds: Deepseek’s cutting-edge model, deployed on Fireworks, with Portkey’s enterprise-grade features:

📊 Full cost & performance insights

⚡ Smart routing between R1 & GPT models

✅ Built-in reliability & safety guardrails

 Link to the cookbook