Together AI pricing
Together AI bills open models in four ways. Serverless inference bills per 1M tokens. Dedicated inference bills per GPU hour. GPU clusters bill per GPU hour. Fine-tuning bills per 1M training tokens. Figures come from the Together AI pricing page on October 27, 2026.
Serverless chat models per 1M tokens
| Model | Input | Cached input | Output |
|---|---|---|---|
| Kimi K3 | $3.00 | $0.30 | $15.00 |
| Qwen3.8-2.4T-A95B | $2.00 | $0.25 | $6.00 |
| GLM-5.3 | $1.40 | $0.26 | $4.40 |
| DeepSeek V4 Pro 0813 | $1.32 | $0.13 | $3.96 |
| Qwen3.7-Max | $1.50 | $0.30 | $4.50 |
| DeepSeek V4.1 Flash | $0.30 | $0.006 | $1.20 |
| GLM-5.3-Flash | $0.15 | $0.03 | $0.50 |
| gpt-oss-120B | $0.15 | None listed | $0.60 |
The page says displayed prices refer to the lowest settings. Image, audio, and video models bill per image, minute, or video. GPT Image 2 costs $0.053 per image. Whisper Large v3 costs $0.0015 per audio minute.
GPU clusters per GPU hour
| GPU | Preemptible | On-demand | Reserved 3 to 30 days | Reserved 91 to 180 days |
|---|---|---|---|---|
| NVIDIA HGX H100 | $1.99 | $3.99 | $3.69 | $3.19 |
| NVIDIA HGX H200 | $2.99 | $5.99 | $4.99 | $3.99 |
| NVIDIA HGX B200 | $4.09 | $8.19 | $7.99 | $6.79 |
| NVIDIA HGX B300 | $4.99 | $9.99 | Contact us | Contact us |
Reserved terms past 181 days are contact-us. GB200 and GB300 NVL72 are contact-us.
Dedicated inference, sandbox, and storage
| Item | Price |
|---|---|
| Dedicated NVIDIA HGX H100 | $5.49 per GPU hour |
| Dedicated NVIDIA HGX B200 | $8.99 per GPU hour |
| Sandbox compute | $0.0446 per vCPU hour, $0.0149 per GiB RAM hour |
| Code Interpreter | $0.03 per 60-minute session |
| Shared filesystem | $0.16 per GiB a month |
| Post-training B200 | $9.00 per GPU hour promo, list $11.00, valid until 12/14/2026 |
Fine-tuning per 1M tokens
| Model | SFT | DPO | Minimum charge |
|---|---|---|---|
| Llama 3.1 8B | $0.34 | $0.84 | $4.00 |
| Llama 3.3 70B | $2.03 | $5.08 | $4.00 |
| gpt-oss-120B | $2.50 | $6.25 | $6.00 |
| Kimi K2.6 | $15.00 | $37.50 | $60.00 |
These rates come from the default LoRA tab. Price is based on training tokens times epochs, plus evaluation tokens. Each job has the minimum charge shown.
What we withhold
We do not list Batch API rates, provisioned throughput, or the cost estimator output. The page shows them per model and per traffic profile.
Compare and track
See the DeepSeek API pricing guide, the Replicate pricing guide, the fal pricing guide, the OpenRouter pricing guide, and all AI tools.
FAQ
How much does Together AI cost per token? It depends on the model. gpt-oss-120B costs $0.15 per 1M input tokens and $0.60 out. Kimi K3 costs $3.00 in and $15.00 out.
How much is an H100 on Together AI? $3.99 per GPU hour on demand in GPU clusters. Preemptible costs $1.99. Dedicated inference on H100 costs $5.49.
Does Together AI have reserved GPU pricing? Yes. H100 costs $3.69 per GPU hour for 3 to 30 days and $3.19 for 91 to 180 days.
How is Together AI fine-tuning priced? Per 1M training tokens, with a minimum charge per job. Llama 3.1 8B LoRA SFT costs $0.34 with a $4.00 minimum.
Is there a Together AI promo? The page lists post-training on B200 at $9.00 per GPU hour, down from $11.00, valid until 12/14/2026.
How we check
We read the Together AI pricing page on October 27, 2026. Get an alert when Together AI changes a price.
Checked 2026-10-27. Prices and policies change. Confirm on the source before you buy.
Sources
More guides
- 1Password pricing per user: Business plan cost (2026)
- ActiveCampaign pricing: plans, seats and contacts (2026)
- Ahrefs pricing per month: Lite, Standard, Advanced, and Enterprise (2026)
- Aircall pricing per user: Essentials, Professional, and the 3-license minimum (2026)
- Airtable pricing per user: who counts as a paid seat, and team totals (2026)
- Amazon Polly pricing: $4 to $100 per million characters (2026)