Together AI adds NVIDIA B200 post-training at $11/GPU/hr, $9 launch promo
What happened
Between the Oct 19, 2026 and Oct 20, 2026 captures of Together AI's pricing page, Pulse recorded 2 changes: New Post-Training on-demand product added to pricing: NVIDIA HGX B200 at $11.00/GPU/hr list, launch promo $9.00/GPU/hr valid until 12/14/2026, billed per minute; New model DeepSeek V4.1 Flash 1017 added to serverless inference pricing at $0.30 input, $0.006 cached, $1.20 output per 1M tokens. Pulse files it as a packaging change (feature added, new plan) with an upward direction. The current page lists Sandbox (Code Interpreter) at $0.03 per session (60 min), Provisioned Throughput at $0.05 per PTU/minute, GPU Clusters (Preemptible) at $1.99 per hour (H100, preemptible), GPU Clusters (Reserved) at $3.69 per hour (H100, 3-30 days) and Sandbox (Code Sandbox) at $0.01 per GiB RAM/hour + $0.04 per vCPU/hour.
It lands one day after Together AI's previous change, "Together AI lists Muse Spark Open, Reflection Beam, Nemotron 3.5 Super VL" (Oct 19, 2026). Before that Pulse recorded "Together AI hikes Qwen3.8 Flash prices up to 68%, adds 10% off Kimi K3 promo" (Oct 12, 2026), "Together AI adds Tev1 4B Experimental at $0.04 input with free output" (Oct 5, 2026) and "Together AI reveals NVIDIA HGX B300 pricing at $9.99/GPU/hr, $4.99 preemptible" (Sep 25, 2026). Together AI has been tracked since 2025-12-23.
In the same window OpenAI, Huggingface, Google, Mistral AI and Anthropic also changed their AI & Machine Learning pricing pages. Those moves are listed further down, and the Pulse Agent can compare them side by side.
- Serverless Inference · FeatureAdded
BeforeNo model DeepSeek V4.1 Flash 1017
Aftermodel DeepSeek V4.1 Flash 1017
- — · PlanAdded
BeforeNot listed
AfterPost-Training on-demand product
| Plan | Field | Before | After | Change |
|---|---|---|---|---|
| Serverless Inference | Feature | No model DeepSeek V4.1 Flash 1017 | model DeepSeek V4.1 Flash 1017 | Added |
| — | Plan | Not listed | Post-Training on-demand product | Added |
Values as written on Together AI's public pricing page in the two captures Pulse compared; where the page states no number, the cells give the wording before and after.
Evidence
The two captures Pulse comparedFull-page captures of Together AI's pricing page (www.together.ai) from Oct 19, 2026 and Oct 20, 2026; the change is marked near the top of the page, where model DeepSeek V4.1 Flash 1017 now appears.
Sign in to see the evidence
The two full-page captures with this change marked up are free with an account.
Together AI, the last 90 days
Full history- Oct 20, 2026improvedTogether AI adds NVIDIA B200 post-training at $11/GPU/hr, $9 launch promo · this change
- Oct 19, 2026improvedTogether AI lists Muse Spark Open, Reflection Beam, Nemotron 3.5 Super VL
- Oct 12, 2026improvedTogether AI hikes Qwen3.8 Flash prices up to 68%, adds 10% off Kimi K3 promo
- Oct 5, 2026improvedTogether AI adds Tev1 4B Experimental at $0.04 input with free output
- Sep 25, 2026packagingTogether AI reveals NVIDIA HGX B300 pricing at $9.99/GPU/hr, $4.99 preemptible
- Sep 23, 2026decreaseTogether AI cuts Qwen3.7-Max and Qwen3.8 Flash inference prices by 40%
- Sep 21, 2026increaseTogether AI: Qwen3.7-Max inference price increased 25% ($2.00→$2.50)
Across AI & Machine Learning in the same window
All AI & Machine Learning changes- OpenAIOct 18OpenAI: Plan taglines reworded to 'Best for...' format: Free (trying out ChatGPT), Go (longer conversations), Plus…
- HuggingfaceOct 16Huggingface adds Nvidia RTX PRO 6000 GPUs to Spaces from $2.75/hr
- GoogleOct 12Google Workspace swaps Gems for AI skills, rewords Gemini on Starter plan
- Mistral AIOct 12Mistral AI cuts Free plan API credits, gives free users Vibe CLI and IDE tools
- AnthropicOct 12Anthropic API launches Haiku 5.5 at $0.10/$0.50 per MTok, Haiku 4.5 goes Legacy
Questions about this change
Did Together AI raise prices in October 2026?
No list-price increase was recorded between Oct 19, 2026 and Oct 20, 2026. The change Pulse detected was a packaging change: feature added, new plan.
What changed on Together AI's pricing page between Oct 19, 2026 and Oct 20, 2026?
Serverless Inference feature: No model DeepSeek V4.1 Flash 1017 → model DeepSeek V4.1 Flash 1017, added; plan: Not listed → Post-Training on-demand product, added. Pulse compared the captures it took on Oct 19, 2026 and Oct 20, 2026; the before/after table above lists each item as written on the page.
How often does Together AI change its pricing?
Pulse recorded 13 other pricing-page changes for Together AI in the 90 days before this one; the previous change was "Together AI lists Muse Spark Open, Reflection Beam, Nemotron 3.5 Super VL" on Oct 19, 2026; Together AI has been tracked since 2025-12-23.