Groq

1 routeUS

Zero data retention

Opt-in

Self-serve toggle in the Console (Data Controls). Enabling ZDR also disables Batch / Fine-tuning / Stored Files.

Training on prompts

No training (third-party open-weight models).

Logging

None for inference by default. Up to 30 days for abuse-monitoring / system-reliability when triggered.

Human review

Abuse-flagged only

GDPR / residency

US-resident; SCCs in Groq Customer DPA

Contract tier

PAYG

Model	Context	Input	Output
GPT OSS 120B OpenAI	128K	$0.15	$0.60
GPT OSS 20B OpenAI	128K	$0.07	$0.30
GPT OSS Safeguard 20B OpenAI	128K	$0.07	$0.30
Llama 3.1 8B Instant Meta	128K	$0.05	$0.08
Llama 3.3 70B Meta	128K	$0.59	$0.79
Llama 4 Scout 17B Meta	128K	$0.11	$0.34
Kimi K2 Instruct 0905 Moonshot	256K	$1.00	$3.00
Kimi K2 Instruct Moonshot	128K	$1.00	$3.00