Skip to content
FrontierDot

Groq free plan limits

The rate-limits page has two tabs, Free Plan Limits and Developer Plan Limits, and two datasets in the page payload named freeRows and devRows. On the HTML fetched for this page, the selected tab was Free Plan Limits and its cells match freeRows. A sentence above the table says the limits shown below are the base limits for the Developer plan. Those two statements do not match. Both tables are copied here, with the labels Groq gives the datasets.

Column tooltips on the page: RPM requests per minute, RPD requests per day, TPM tokens per minute, TPD tokens per day, ASH audio seconds per hour, ASD audio seconds per day. A dash in the payload is shown here as an em dash. Rate limits apply at the organization level. Cached tokens do not count towards rate limits. The page says the table is a high-level summary and that the exact limits for an organization are on the account limits page.

Free Plan Limits (freeRows)

ModelRPMRPDTPMTPDASHASD
canopylabs/orpheus-arabic-saudi101001.2K3.6K——
canopylabs/orpheus-v1-english101001.2K3.6K——
meta-llama/llama-prompt-guard-2-22m3014.4K15K500K——
meta-llama/llama-prompt-guard-2-86m3014.4K15K500K——
openai/gpt-oss-120b301K8K200K——
openai/gpt-oss-20b301K8K200K——
openai/gpt-oss-safeguard-20b31K2K200K——
qwen/qwen3.8-27b301K8K200K——
whisper-large-v3202K——7.2K28.8K
whisper-large-v3-turbo202K——7.2K28.8K

Developer Plan Limits (devRows)

These are the higher figures in devRows, not the free plan. The models page labels a matching column “RATE LIMITS (DEVELOPER PLAN)”. For openai/gpt-oss-120b that models-page cell is 250K TPM and 1K RPM, which matches devRows.

ModelRPMRPDTPMTPDASHASD
canopylabs/orpheus-arabic-saudi250100K50K———
canopylabs/orpheus-v1-english250100K50K———
meta-llama/llama-prompt-guard-2-22m10050K30K———
meta-llama/llama-prompt-guard-2-86m10050K30K———
openai/gpt-oss-120b1K500K250K———
openai/gpt-oss-20b1K500K250K———
openai/gpt-oss-safeguard-20b1K500K150K———
qwen/qwen3.8-27b1K500K250K———
whisper-large-v3300200K——200K4M
whisper-large-v3-turbo400200K——400K4M

What the free plan does not include

The rate-limits page tells you to upgrade to the Developer plan for higher limits, Batch, and Flex processing. The services agreement says some Cloud Services and AI Model Services may be designated fee-free for a limited time or based on usage limits. It does not list which models are fee-free.

Data use

Services agreement section 4.2: Groq does not access, use, store, or retain Inputs or Outputs except as necessary to provide the Cloud Services, in accordance with the customer’s permission or instruction, to comply with law, to ensure reliable operation, or to confirm compliance with the AUP. Eligible customers may turn on a zero data retention setting in the console. Groq is not permitted to use Inputs or Outputs for training or fine-tuning unless the customer explicitly grants permission. The agreement does not define “eligible customers” in that sentence, and it does not split the rule by free versus paid plan.

How to get a key

Create a key at the Groq console keys page and set GROQ_API_KEY. The quickstart points at that keys URL.

Other free-tier notes

Groq in the directory

Sources

Last checked 2026-10-22.