Model APIs - 14 entries
Groq API rate limits
Groq's official docs, read on October 17, 2026, set rate limits for the Groq API across Developer plan, Paid customers only and Paid tier. For canopylabs/orpheus-arabic-saudi on Developer plan (base limits, as labelled on the page), the docs list 10 RPM, 100 RPD and 1.2K TPM.
Last verified from console.groq.com. What the docs leave out
- Tiers
- 3
- Tables
- 6
- Entries
- 14
- Verified
- Oct 17, 2026
Limits by model, endpoint and plan
Values are copied exactly as the docs show them, units included, in the docs' own order. A dash means the docs list no value for that row.
All 6 tables on this page
canopylabs/orpheus-arabic-saudi, canopylabs/orpheus-v1-english and 6 more
8 scopes - Developer plan (base limits, as labelled on the page)
| Model or scope | RPM |
|---|---|
| canopylabs/orpheus-arabic-saudi | 10 |
| |
| canopylabs/orpheus-v1-english | 10 |
| |
| meta-llama/llama-prompt-guard-2-22m | 30 |
| |
| meta-llama/llama-prompt-guard-2-86m | 30 |
| |
| openai/gpt-oss-120b | 30 |
| |
| openai/gpt-oss-20b | 30 |
| |
| openai/gpt-oss-safeguard-20b | 3 |
| |
| qwen/qwen3.8-27b | 30 |
| |
- canopylabs/orpheus-arabic-saudi: Groq publishes one table only, with no per-plan tabs. The page wording is ambiguous: it says to upgrade to the Developer plan for higher limits, then says the table shows the base limits for the Developer plan. Groq calls it "a high level summary and there may be exceptions to these limits". Each organization's exact limits are on the logged-in Limits page (console.groq.com/settings/limits). Limits apply per organization, not per user. Cached tokens do not count toward rate limits.
- canopylabs/orpheus-v1-english: Text-to-speech model. The same single-table caveat applies as above.
- openai/gpt-oss-120b: Some organizations also have separate ITPM/OTPM caps. They show on the Limits page as an "X in / Y out" breakdown when you hover over TPM.
- qwen/qwen3.8-27b: The model ID is copied verbatim from the table.
whisper-large-v3 and whisper-large-v3-turbo
2 scopes - Developer plan (base limits, as labelled on the page)
| Model or scope | RPM |
|---|---|
| whisper-large-v3 | 20 |
| |
| whisper-large-v3-turbo | 20 |
| |
- whisper-large-v3: Speech-to-text model, limited by audio seconds instead of tokens.
All models (organization level) - rate limit headers and 429 behavior
| Metric | Limit |
|---|---|
| retry-after | 2 (In seconds; illustrative) |
| x-ratelimit-limit-requests | 14400 (Always refers to Requests Per Day (RPD); illustrative) |
| x-ratelimit-limit-tokens | 18000 (Always refers to Tokens Per Minute (TPM); illustrative) |
| x-ratelimit-remaining-requests | 14370 (RPD; illustrative) |
| x-ratelimit-remaining-tokens | 17997 (TPM; illustrative) |
| x-ratelimit-reset-requests | 2m59.56s (RPD; illustrative) |
| x-ratelimit-reset-tokens | 7.66s (TPM; illustrative) |
- When you exceed a limit the API returns "429 Too Many Requests". "retry-after is only set if you hit the rate limit and status code 429 is returned. The other headers are always included." Limits apply at the organization level, and you hit whichever threshold you reach first. Metrics defined on the page are RPM, RPD, TPM, TPD, ASH, ASD, ITPM and OTPM.
All models - Flex processing (service tier)
Paid customers only
| Metric | Paid customers only |
|---|---|
| Rate limit multiplier | 10x higher rate limits compared to on-demand processing |
- "Requests run at higher rate limits while capacity is available." "If flex capacity is unavailable, requests will fail quickly with status 498 and error capacity_exceeded."
Batch API
| Metric | Limit |
|---|---|
| Batch input file size | .jsonl files 50,000 lines or less and up to maximum of 200MB in size |
| Number of batch jobs | There is no limit for the number of batch jobs you can submit. |
| Completion window | 24 hours (24h) to 7 days (7d) |
- "The Batch API rate limits are separate than existing per-model rate limits for synchronous requests. Using the Batch API will not consume tokens from your standard per-model limits." Batch-specific numeric rate limits are only shown in the logged-in console.
Organization-wide monthly spend limit (user-configured)
Paid tier
| Metric | Paid tier |
|---|---|
| Spend limit window | Monthly; resets on the 1st of each month |
| Spend tracking delay | 10-15 minutes |
- This is a budget cap you set yourself, not a Groq rate limit. When you hit it, "API calls from any key in your organization will return a 400 with code blocked_api_access". Only paid plans can set one, and you need organization owner permissions.
How to raise your limits
canopylabs/orpheus-arabic-saudi
"Need higher rate limits? Upgrade to Developer plan to access higher limits, Batch and Flex processing, and more. Note that the limits shown below are the base limits for the Developer plan, and higher limits are available for select workloads and enterprise use cases."
canopylabs/orpheus-v1-english, meta-llama/llama-prompt-guard-2-22m and 7 more
Upgrade to Developer plan (console.groq.com/settings/billing/plans). Higher limits are available for select workloads and enterprise use cases.
All models - Flex processing (service tier)
Use Flex processing on a paid plan.
Organization-wide monthly spend limit (user-configured)
You can raise or remove the spending limit at any time under Settings -> Billing -> Limits to restore access immediately.
What the docs leave out
- Separate per-plan tables: Groq publishes one table and does not split it into Free and Developer plans. The page tells readers to upgrade to the Developer plan for higher limits, then labels the same table as the Developer plan base limits, so which plan those numbers belong to is ambiguous.
- Exact per-organization limits, including any ITPM/OTPM breakdown, are only visible on the logged-in Limits page (console.groq.com/settings/limits).
- Batch API numeric rate limits: the docs say they are separate from synchronous limits but only link to the logged-in console for the values.
- Enterprise or 'select workloads' higher limits: not published, and no request form or criteria are documented.
- Free-tier specifics and Developer plan pricing or eligibility: groq.com/pricing redirected to the homepage, and the plans page (console.groq.com/settings/billing/plans) requires login.
- Rate limits for models not in the table (for example Llama chat models or Kimi): not listed on the rate-limits page today.
FAQ
What are the Groq API rate limits?
Groq's official docs, read on October 17, 2026, set rate limits for the Groq API across Developer plan, Paid customers only and Paid tier. For canopylabs/orpheus-arabic-saudi on Developer plan (base limits, as labelled on the page), the docs list 10 RPM, 100 RPD and 1.2K TPM.
How do I raise my Groq API rate limits?
"Need higher rate limits? Upgrade to Developer plan to access higher limits, Batch and Flex processing, and more. Note that the limits shown below are the base limits for the Developer plan, and higher limits are available for select workloads and enterprise use cases." Upgrade to Developer plan (console.groq.com/settings/billing/plans). Higher limits are available for select workloads and enterprise use cases. Use Flex processing on a paid plan.
Where do these Groq numbers come from?
Every value is copied verbatim from Groq's official documentation at console.groq.com, read on October 17, 2026. This page lists all 14 entries in 6 tables, each linked to the page it came from.
Sources
Last verified . Every value is copied verbatim from the official pages below, read on that date.
Newsletter
Know when limits change
One email per week. New model releases, pricing and limit changes, and what they mean for what you build.