Model APIs - 2 entries
Together AI rate limits
Together AI's official docs, read on October 17, 2026, set rate limits for Together AI. For Provisioned throughput (GLM-5.3, GLM-5.3 Flash, DeepSeek V4.1 Flash, GLM-5.2, Kimi K3) on Provisioned throughput (PTU commitment), the docs list PTU price: $0.05 per PTU per minute; Minimum term: one month.
Last verified from docs.together.ai. What the docs leave out
- Sources
- 2
- Tables
- 2
- Entries
- 2
- 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.
Serverless inference under Enterprise or Scale contracts
Enterprise / Scale contract
| Metric | Enterprise / Scale contract |
|---|---|
| Purchased rate limits | "If you have an active Enterprise or Scale contract, your purchased rate limits stay in place until the contract expires. Nothing changes during your current term." |
Provisioned throughput (GLM-5.3, GLM-5.3 Flash, DeepSeek V4.1 Flash, GLM-5.2, Kimi K3)
Provisioned throughput (PTU commitment)
| Metric | Provisioned throughput (PTU commitment) |
|---|---|
| TPM | Throughput up to the maximum your purchased PTUs entitle you to for the selected model or model family, measured in TPM; TPM per PTU per token type is model-specific and published on the pricing page |
| PTU price | $0.05 per PTU per minute |
| Minimum term | one month |
| Reliability SLA | At least 99% of eligible requests complete successfully each month |
- Overage: "Traffic above your purchased PTU capacity may fall back to the shared serverless pool on a best-effort basis, billed at standard serverless rates" and may receive 429 or 503. Unused PTUs don't roll over.
How to raise your limits
Serverless inference under Enterprise or Scale contracts
Per contract; contact sales.
Provisioned throughput (GLM-5.3, GLM-5.3 Flash, DeepSeek V4.1 Flash, GLM-5.2, Kimi K3)
Buy more PTUs through sales ("Provisioned throughput is currently unavailable for self-service"); "If you consistently run above your PTU commitment, you may be able to expand it through your sales contact."
What the docs leave out
- Together AI no longer publishes numeric serverless rate limits (RPM, TPM, RPD, concurrency) or a usage tier table on its rate-limits page as of 2026-10-17. Serverless limiting is described as dynamic and demand-based. No tier advancement rules are documented.
- Per-model serverless limits are not published. The docs say only that Together "may apply custom rate limits or access restrictions" to high-demand models.
- The TPM per PTU values for provisioned throughput are on the pricing page (together.ai/pricing#provisioned-throughput), not in the docs. I did not fetch them.
- Contracted rate limits for Enterprise and Scale customers and dedicated endpoint GPU quota values are not public.
- Fine-tuning and post-training model limits (context length, batch size) are documented separately (docs/fine-tuning/supported-models, reference/get-fine-tunes-models-limits). I did not fetch them because they are training limits, not rate limits.
FAQ
What are the Together AI rate limits?
Together AI's official docs, read on October 17, 2026, set rate limits for Together AI. For Provisioned throughput (GLM-5.3, GLM-5.3 Flash, DeepSeek V4.1 Flash, GLM-5.2, Kimi K3) on Provisioned throughput (PTU commitment), the docs list PTU price: $0.05 per PTU per minute; Minimum term: one month.
How do I raise my Together AI rate limits?
Per contract; contact sales. Buy more PTUs through sales ("Provisioned throughput is currently unavailable for self-service"); "If you consistently run above your PTU commitment, you may be able to expand it through your sales contact."
Where do these Together AI numbers come from?
Every value is copied verbatim from Together AI's official documentation at docs.together.ai, read on October 17, 2026. This page lists all 2 entries in 2 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.