We may earn a commission if you buy through links on this page. It never changes a score or a ranking. How we make money
There's a free plan; paid tiers are quoted.
Together AI lists a free plan, with some model options priced at $0 per 1M tokens. The vendor does not publish the free plan’s allowance, so check the account limits before using it for a sustained workload. Paid inference is billed by token, while reserved capacity is priced per PTU per minute and dedicated infrastructure per GPU-hour. Media generation has separate per-image, per-video and audio rates; there is no single monthly subscription price for the platform.
01Every plan, line by line
| Plan | Billed annually | Billed monthly | Built for |
|---|---|---|---|
| Serverless Inference | — | Quote | Teams serving models on demand |
| Batch Inference | — | Quote | Teams processing batch workloads |
| Provisioned Throughput | — | Quote | Production teams reserving capacity |
| Dedicated Inference | — | Quote | Teams needing dedicated GPU infrastructure |
| Dedicated Container Inference | — | Quote | Teams deploying generative media models |
Add-ons and overages
- Image generation examples: $0.0154 per image for FLUX.2 dev; $0.053 per image for GPT Image 2.
- Video generation examples: $0.115 per video for ByteDance Seedance 2.5; $0.57 per video for ByteDance Seedance 1.0 Pro.
- Audio examples: $0.0015 per audio minute for Whisper Large v3; $65 per 1M characters for Cartesia Sonic-3.
- Displayed image prices use listed default steps; additional costs apply when exceeding default steps. Actual prices may vary by settings.
02What the plan grid doesn't tell you
Output token rates
Kimi K3 costs $3 per 1M input tokens and $15 per 1M output tokens, so output-heavy workloads can cost more than input volume suggests.
Reserved capacity
Provisioned Throughput is priced per PTU per minute. The estimator assumes about 43,800 minutes monthly, so confirm the capacity and billing commitment for your workload.
GPU quotes
HGX H200, B300 and GB200 NVL72 rates, along with reserved GPU rates, require a quote. Dedicated Container Inference also uses custom pricing.
Media settings
Image prices use listed default steps. Additional costs apply beyond those steps, and actual prices may vary by settings.
03Which plan should you buy?
| If you are… | Get | Why |
|---|---|---|
| Side project testing models | Serverless Inference | Start with the free plan or free model options; pay per token for models with listed rates. |
| Team processing large datasets | Batch Inference | The batch option handles workloads up to 30 billion tokens per model, with model-specific token rates. |
| Production team reserving capacity | Provisioned Throughput | Choose token-based reserved capacity with a listed 99% uptime SLA; confirm PTU costs for the required capacity. |
04Is Together AI worth the money?
Together AI can be inexpensive for experimentation when a suitable free model or low-cost token rate fits the workload. For production, model choice, output-token volume and reserved capacity determine the bill; dedicated GPUs and custom container deployments add costs that may require a quote. There is no published monthly subscription price to use as a simple budget baseline. Start with the free plan for evaluation, then estimate against expected input and output tokens or GPU hours. Get written pricing for custom capacity before committing.
Together AI pricing FAQ
Is Together AI free to use?
Together AI lists a free plan, and some model options have $0 token prices. Ternary Bonsai 27B is listed at $0 per 1M input and output tokens. The free plan’s allowance is not published, so check the limits attached to your account.
What is the cheapest listed Together AI model?
Ternary Bonsai 27B is listed at $0 per 1M input and output tokens. Tev1 4B Experimental is listed at $0.04 per 1M input tokens with free output. Prices and model availability can change, so check the current model pricing before deployment.
How much does Together AI charge for GPU inference?
Dedicated Inference lists HGX H100 on-demand at $5.49 per GPU-hour and HGX B200 at $8.99 per GPU-hour. HGX H200, B300 and GB200 NVL72 rates are available by contacting sales. Reserved rates also require a sales quote.
Does Together AI charge a monthly subscription?
The listed pricing is usage-based, covering token consumption, PTU minutes, GPU hours and media generation. The pricing page lists no monthly or annual subscription billing. Provisioned Throughput uses per-PTU-per-minute pricing, and the estimator assumes about 43,800 minutes monthly.