
Together AI
San Francisco, CA-based AI infrastructure company founded 2022. Private company. Operates in AI cloud computing, model training, and inference. Serves developers, enterprises, and researchers with open-source AI models and decentralized cloud services.
Pricing Change History
47 changes · 10 summariesAdded Qwen4 Max model to Serverless Inference (Chat) pricing at $1.43/1M input tokens, $4.28/1M output tokens, $0.0713/1M cached input tokens
Serverless Inference chat-model catalog updated: added Elorian Glimpse 1.0 ($0.60/1M input, $3.00/1M output, $0.06/1M cached); removed Rnj-1 Instruct ($0.15/$0.15) and Qwen2.5 7B Instruct Turbo ($0.30/$0.30) from the chat pricing table.
GPU hardware rates rose across Dedicated Inference & GPU Clusters (HGX B200 $8.99→$9.29/hr); previously-hidden H200, B300 and GB300 NVL72 prices are now publicly shown.
Includes Nov 13, Nov 11 · first seen Nov 11, 2026
Added US geo-pinned model variants (e.g. Kimi K3 US at $3.75/$18.75) and live GPU capacity status (Available/Sold Out) with Request Capacity buttons.
Qwen3.7-Max rose $1.50/$4.50 to $2.50/$7.50 per 1M tokens; Kimi K3.1 Pro and Ripple-65B Preview added; Qwen3.8-2.4T, Inkling, Qwen3.5-397B, Cogito, Llama 3 8B Lite, Qwen3 235B removed.
Includes Nov 7, Nov 5 · first seen Nov 5, 2026
Qwen3.7-Max serverless inference pricing increased from $1.50/$0.30 (cached)/$4.50 to $2.50/$0.50 (cached)/$7.50 per 1M input/cached/output tokens (67% across the board); Inkling, Cogito v2.1 671B, Llama 3 8B Instruct Lite, and Qwen3 235B A22B Instruct 2507 FP8 Throughput were removed from the Serverless Inference model lineup
Added two new models to serverless inference: Kimi K3.1 Pro ($2.00/$10.00 per 1M input/output tokens, $0.20 cached) and Ripple-65B Preview ($0.20/$0.60 per 1M tokens, $0.02 cached)
Four serverless inference models dropped from the pricing list: Gemma 4 31B, DeepSeek V4 Flash 0731, Llama 3.3 70B and MiniMax M2.7.
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 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
Limited-time free promo for Muse Spark Open (was $0.00 in/out, valid until 10/18/2026) removed after expiry; standard $1.25 input / $4.25 output pricing remains.
New serverless models listed: Muse Spark Open ($1.25 in/$4.25 out), Reflection Beam ($0.40/$1.20) and NVIDIA Nemotron 3.5 Super VL ($0.15/$0.50) per 1M tokens.
Includes 4 dailies · first seen Oct 14, 2026
New model Reflection Beam added to serverless inference pricing at $0.40 input ($0.04 cached) and $1.20 output per 1M tokens
Muse Spark Open promo row added: input $1.00 and output $5.00 become $0.00 until 10/18/2026; site banner offers free trial through 10/18 12 p.m. PT Serverless model rate card adds Muse Spark Open ($1.25 in / $4.25 out, $0.15 cached) and NVIDIA Nemotron 3.5 Super VL ($0.15 in / $0.50 out, $0.02 cached)
Kimi K3 cached input token price in the Serverless Inference (Chat) pricing table rose from $0.27 to $0.30 per 1M tokens; standard input $3.00 and output $15.00 unchanged.
Qwen3.8 Flash serverless price up from $0.09 input/$0.28 output to $0.15/$0.47 per 1M tokens. Kimi K3 PROMO pricing added: $2.70 input/$13.50 output (from $3.00/$15.00) per 1M tokens, valid until 10/11/2026.
Includes Oct 8, Oct 6 · first seen Oct 6, 2026
GPU Clusters reserved pricing first term bracket changed from 7-30 days to 3-30 days (shorter minimum reservation); prices unchanged.
Serverless token prices changed: Kimi K3 cut to $2.70 in / $13.50 out per 1M tokens (from $3.00 / $15.00); Qwen3.8 Flash raised to $0.15 / $0.47 (from $0.09 / $0.28)
New serverless model Tev1 4B Experimental added to model pricing table at $0.04 input with a Free output tier.
Includes Oct 2 · first seen Oct 2, 2026
Added Tev1 4B Experimental model to serverless pricing table at $0.04 input, Free output
Current plans
| Plan | Price |
|---|---|
| Provisioned Throughput | $0.05 per PTU/minute |
| GPU Clusters (On-demand) | $3.99 per hour (H100) |
| GPU Clusters (Reserved) | $3.69 per hour (H100, 3-30 days) |
| Sandbox (Code Sandbox) | Usage-based |
Also priced by usage
| |
| Serverless Inference | $0.15 per 1M input tokens |
| Fine-Tuning | Usage-based |
Also priced by usage
| |
| GPU Clusters (Preemptible) | $1.99 per hour (H100, preemptible) |
| Managed Storage | $0.16 per GiB/month |
| Post-Training | $9 per GPU‑hour (B200, promo) |
| |
| Dedicated Inference | Usage-based |
Also priced by usage
| |
| Sandbox (Code Interpreter) | $0.03 per session (60 min) |
Company Facts
Pulse news about Together AI
View all →











How Together AI compares with its peers
| Company | Entry price |
|---|---|
A AnthropicLast change: Nov 16, 2026![]() | $17 /mobilled annually |
O OpenAILast change: Nov 13, 2026![]() | $8 /mo |
G GoogleLast change: Oct 12, 2026 | Contact |
M Mistral AILast change: Oct 12, 2026![]() | $14.99 /mo |
H HuggingfaceLast change: Nov 17, 2026![]() | $9 /mo |
L Lambda LabsLast change: Nov 9, 2026![]() | $0.69 per GPU per hour |
Together AI pricing FAQ
- How much does Together AI cost?
- Together AI prices on usage: the Provisioned Throughput plan is $0.05 per PTU/minute.
- Does Together AI have a free plan?
- No. Together AI does not list a free tier.
- What pricing model does Together AI use?
- Together AI uses usage-based pricing across 11 plans.
- When did Together AI last change its pricing?
- Nov 19, 2026: Together AI adds Qwen4 Max model at $1.43/$4.28 per 1M tokens.
- How often has Together AI changed its pricing?
- Pulse has recorded 47 pricing changes for Together AI since it started tracking on Dec 23, 2025; the pricing page was last checked on Nov 19, 2026.
- What does “tracked since” mean?
- It's the date of the first saved copy of the pricing page that Pulse checked for changes against a later copy. For a few companies we hold older copies that Pulse never checked for changes, and those don't change this date. If Pulse has never checked two copies, it's the date of the first saved copy.
- Who are Together AI's pricing peers?
- Pulse compares Together AI with Anthropic, OpenAI, Google, Mistral AI, Huggingface and Lambda Labs.




