April Cool: 154 PR Merges
I sat down to count our PRs for the last month - 154 PRs merged, resulting in a total of 30 distinct updates. (that's 1 every day). I think we can do more! But we'll compensate for that with what we are shipping this month. 👀
In the meanwhile, here's a rundown of everything we shipped at Portkey last month:
New Features
Update Feedback: Feedbacks now have distinct IDs, allowing you to update any previous feedback object. (more)
Saved Filter: You can now save frequently used filter combinations and access them directly from the Logs or Analytics search bar. (more)
Do Not Track: Process requests without logging the request and response data. (more)
Templating Engine: Make your prompt templates more extensible with MustacheJS tags. (more)
Okta SSO: Portkey supports Okta SSO for Enterprises now.
New Meetups
Ayush & I are in SF this month and would love to catch you at RSA or in Seattle, or for a filter coffee somewhere around SoMa? (just send a hi at rohit [at] portkey.ai)
We also did the first LLMs in Prod meetup in NYC last month. Really great food, even better conversations. One Portkey user flew down from Spain just for the event.
New 98% Cost Savings
We wrote a mega implementation guide for FrugalGPT that can get you upto 98% cost savings on LLMs without compromising on accuracy.
New Partnerships
Portkey Goes Enterprise (and how!)
F5 (the creators of NGINX) is partnering with Portkey to bring our observability suite, AI gateway, and prompts platform to the F5 Distributed Cloud.
Read the announcement here. Schedule a call here if you'd like to talk more.
What Else?
- Llama 3 is available on prompt playground and with 9+ providers
- Deeper docs for Fireworks, Groq, Jina AI, Bedrock, Ollama
- You can now bring your own custom LLM to Portkey
- Get observability (and more) for your Vercel AI app with Portkey
- Our Langchain documentation is revamped
- Use Portkey as a prompt hub for your promptfoo evals (very cool)
- Cool to see an open source project powered by Portkey Gateway
That's it! Want us to prioritize a feature or an integration this month? Join the LLMs in Prod Community below to share your feedback AND to meet leading practitioners wrangling with LLMs every day
