Together AI: Together AI's October 16, 2026 changelog entry for geo-specific inference no longer lists Canada (https://ca-api.together.ai/v1) as an option; it now lists the United States only, at the same 1.25× rate.
Old "You can now keep serverless inference processing in the United States or Canada." Canada base URL https://ca-api.together.ai/v1 with models zai-org/GLM-5.3, zai-org/GLM-5.3-Flash, zai-org/GLM-5.2, deepseek-ai/DeepSeek-V4.1-Flash, deepseek-ai/DeepSeek-V4-Pro-0813. "Both geographies cost 1.25× standard serverless rates (a 25% premium) for input, cached input, and output tokens."
New "You can now keep serverless inference processing in the United States." Canada: not stated. "US-only inference costs 1.25× standard serverless rates (a 25% premium) for input, cached input, and output tokens." Reason for the change: not stated. Effective date: not stated.
Why it matters: If you are on or choosing a Together API (geo-specific serverless inference) plan, this changes what the plan includes or how it is billed. The source does not state a separate effective date.
Alerts waitlist (not launched)
Joining sends nothing now; we keep your email only to tell you when alerts launch. Remove it any time: support@testedmcp.com. Stored by Netlify Forms. Privacy
| Product | Together API (geo-specific serverless inference) |
|---|---|
| Category | plan |
| Announced / detected | 2026-10-28 |
| Effective | not stated in source |
| Before | "You can now keep serverless inference processing in the United States or Canada." Canada base URL https://ca-api.together.ai/v1 with models zai-org/GLM-5.3, zai-org/GLM-5.3-Flash, zai-org/GLM-5.2, deepseek-ai/DeepSeek-V4.1-Flash, deepseek-ai/DeepSeek-V4-Pro-0813. "Both geographies cost 1.25× standard serverless rates (a 25% premium) for input, cached input, and output tokens." |
| After | "You can now keep serverless inference processing in the United States." Canada: not stated. "US-only inference costs 1.25× standard serverless rates (a 25% premium) for input, cached input, and output tokens." Reason for the change: not stated. Effective date: not stated. |
| Official source | https://docs.together.ai/docs/changelog |
| Source quote | US-only inference costs 1.25× standard serverless rates (a 25% premium) for input, cached input, and output tokens. |
| Archive | No post-change Wayback capture found |
| Snapshot hash | sha256:45d125bd47cc2b7876d240ac05795e35272132d23b64dd29d613e2350ee0ab6b |
| Status | verified |
All Together AI changes · Dispute or correct this entry
Previous changes from Together AI
- 2026-10-02: Dynamic serverless rate limits that scaled with recent traffic were retired.
- 2026-09-22: Serverless prices lowered for Qwen3.7-Max and Qwen3.8-Flash.
- 2026-06-09 Backfilled: DeepSeek-V4-Pro serverless price lowered; cached-input pricing added for GLM-5.1 and Qwen3.5-397B-A17B.
- 2026-05-29 Backfilled: Serverless prices increased for Qwen3.5-9B, Meta-Llama-3-8B-Instruct-Lite and Llama-3.3-70B-Instruct-Turbo.
- 2026-05-14 Backfilled: Announced price increase for google/gemma-4-31b-it effective 2026-05-21.
- 2026-04-24 Backfilled: Billable training tokens for no-packing fine-tuning jobs now calculated as len(dataset) * max_seq_length.
- 2026-04-22 Backfilled: Build Tiers 1-5, Scale and Enterprise tier labels retired, dynamic rate limits enabled for all users, billing moved to fully prepaid.
- 2026-04-15 Backfilled: Serverless price increased for google/gemma-3n-E4B-it.
How Together AI limits and pricing work · Together AI changes, October 2026