Why choose Felafax Gateway?
Controlled prompt change rollouts
Introduce "shadow" prompts and run them on 1% of live traffic alongside your main prompt. Roll out the new prompt to more users as you gain confidence, minimizing risk and iterating rapidly with prompt tweaks.
Switch between GPT4o and Claude3.5
We've built a high-performance Rust-based LLM router with zero latency. Integrate with one line of change to easily switch between different LLM models.
Live eval with "shadow" traffic
No need to painstakingly create eval datasets. Fork production traffic to test multiple prompt versions simultaneously. We also have built-in metrics like helpfulness and hallucination.
Semantic Search
We log all your LLM API requests/responses for enhanced observability and provide powerful semantic search through your logs.
Continuous Evals
Configure a subset of your traffic (like 1%) and run out-of-the-box eval metrics to continuously monitor your app. No need to rely on "vibe" checks for your LLM setups.
Onboard with one line
Integrate Felafax Gateway by simply changing the base_url in your OpenAI client. This one-line modification gives you instant access to our advanced LLM management features without disrupting your existing workflow.
How our Gateway works?
01
Set base_url in OpenAI python API
Point your OpenAI API to the Felafax Proxy with a simple base_url change.
02
View live traffic on our dashboard
Instantly see API requests and responses live on our dashboard after changing the base_url.
03
Rollout new prompt changes with "shadow" prompts
Introduce "shadow" prompts in your Python code and use our dashboard to test them with 1% of your users!
FAQ