Groq integration
Mask personal data before it reaches Groq
- Every text field in the request body: user and system messages, tool and function-call arguments, and tool results
- Streaming responses
- Base64, hex, and percent-encoded text inside the request is decoded and checked too
Teams pick Groq for speed, so any control in front of it has to be fast too. Maskflare's detection engine is deterministic: patterns, checksums, and dictionaries rather than a second model, so it adds milliseconds, not another model call.
Requests go through your gateway, sensitive values are masked before they reach Groq, and with restoration on, the answer comes back with the real values.
Set up Groq with Maskflare
Requests go to /v1/groq on your Maskflare gateway instead of the provider.
from openai import OpenAI
client = OpenAI(
base_url="https://YOUR_GATEWAY_HOST/v1/groq",
api_key="mf_live_...", # a Maskflare key
)
reply = client.chat.completions.create(
model="llama-3.3-70b-versatile",
messages=[{"role": "user", "content": "Route this message from +44 20 7946 0958"}],
)What gets masked
- Every text field in the request body: user and system messages, tool and function-call arguments, and tool results
- Streaming responses
- Base64, hex, and percent-encoded text inside the request is decoded and checked too
Good to know
- In our engine benchmarks, masking a 10 KB chat request with the default rules takes about 2 ms (median).
- A rule set to block stops the request before it reaches the provider and returns HTTP 403 with the matching rule keys, never the values.
- Store the provider key once in the Maskflare console, or keep sending it per request in pass-through mode.
- No code change option: the Maskflare forward proxy inspects traffic to api.groq.com with the same rules.
Frequently asked questions
How much latency does masking add?
Maskflare's engine benchmarks show a median of about 2 ms to mask a 10 KB chat request with the default rules. Network time to the gateway comes on top and depends on where you call it from.
Does Maskflare use an AI model to find PII?
No. Detection is deterministic: validated patterns, checksums, dictionaries, and Exact Data Match. The same input always gives the same result.
Which Groq endpoints are covered?
Every path under /v1/groq is forwarded to Groq's OpenAI-compatible API with the request body masked.
Other providers
See it on your own data
Book a 30-minute demo.
Bring your hardest prompts.
We'll show detection, masking, and restoration on your providers and data types,
and how it fits your stack.