Together AI Pricing Calculator
Updated July 2026Pure-usage per-token serverless + per-hour dedicated GPU + reserved capacity + sales-led enterprise
Together AI pricing: Together AI pricing 2026: per-token serverless (Llama, DeepSeek, FLUX), H100 $5.49 dedicated, $3.59 reserved cluster, B200 $8.99, PTU throughput, Batch 50%. Use the free calculator below to enter your usage and get an instant all-in monthly estimate — including overages — and see which plan is cheapest for your needs.
How Together AI prices: Pure-usage per-token serverless + per-hour dedicated GPU + reserved capacity + sales-led enterprise.
How Together AI prices
Pure-usage per-token serverless + per-hour dedicated GPU + reserved capacity + sales-led enterprise
Your usage
💡 1M input tokens ≈ 750k words of prompts + context; a busy RAG app sends 100M+/mo
Together publishes per-model serverless rates inline. Pick the model you'll run; input and output are billed per 1M tokens at the model's rate.
💡 Output tokens are the model's generated response — usually a fraction of input for RAG, more for agents
Match this to the same model you picked for the input rate — output tokens are usually priced higher than input.
💡 Image SKUs bill per generated image; leave at 0 if you only run text models
Together prices most image models per generated image at the default resolution and step count.
💡 A dedicated endpoint running 24×7 = ~730 GPU-hours/mo; leave at 0 for pure serverless
Dedicated endpoints are single-tenant; clusters are multi-node. Reserved cluster rates require a 7–30 day commit.
💡 Agentic code execution bills per vCPU-hour and per GiB-hour separately
💡 Memory attached to sandbox runtime, billed per GiB-hour alongside vCPU-hours
Monthly Estimate
Pay-as-you-go
$135.20
per month
Cost Breakdown
Annual
$1,622.40
Why Pay-as-you-go — $135.20/mo
- Pay-as-you-go at $135.20/mo.
- How we got there: 100 M tokens (LLM input tokens) × 1 $/1M in (Chat model (input rate)) = 104 $.
- How we got there: 30 M tokens (LLM output tokens) × 1 $/1M out (Chat model (output rate)) = 31 $.
- How we got there: 0 images (Generated images) × 0 $/image (Image model (per-image rate)) = 0 $.
- How we got there: 0 GPU-hours (Dedicated / cluster GPU hours) × 5 $/hr (GPU type (per-hour rate)) = 0 $.
- LLM input tokens (104 $ × $1) is 77% of your bill — the lever that matters most.
- Need less? Chat model (input rate): gpt-oss-120B — $0.15/1M in drops you to Pay-as-you-go at $46.20/mo.
Budget range
Plan for usage swings, not just today's estimate.
Conservative
$83.20
50 M tokens/mo
Expected
$135.20
100 M tokens/mo
Aggressive
$239.20
200 M tokens/mo
Need a calculator like this on your pricing page?
Embed interactive pricing calculators on your website to help customers understand costs and boost conversions.
Get StartedAbout this Together AI calculator
This calculator estimates your Together AI cost from publicly available pricing. Actual costs may vary with your specific agreement, volume discounts, and usage patterns — always verify on the provider's official pricing page for the most current rates.
Together AI pricing — frequently asked questions
How much does Together AI cost per month? ▼
Together has no monthly subscription fee — you pay only for the serverless tokens, dedicated GPU hours, fine-tuning training tokens, and Code Sandbox usage you consume. A small RAG application using Llama 3.3 70B at 30M input + 10M output tokens would cost ~$42/month on serverless; the same workload on a dedicated H100 ($5.49/hr) running 4h/day would cost ~$660/month.
What are Together's serverless per-token rates? ▼
Together publishes per-model rates inline on the pricing page. Sample rates per 1M tokens: Llama 3.3 70B at $1.04 input / $1.04 output; DeepSeek V4 Pro at $1.74 input / $3.48 output ($0.20 cached input); Qwen3.5 9B at $0.17 input / $0.25 output; GLM-5.1 at $1.40 input / $4.40 output. Image generation: FLUX.2 [dev] at $0.0154/image, FLUX.1 [schnell] at $0.0027/image, Stable Diffusion 3 at $0.0019/image.
What are Together's GPU rates for dedicated endpoints and clusters? ▼
Dedicated inference (per GPU per hour, on-demand): HGX H100 at $5.49/hr, HGX B200 at $8.99/hr (H200/B300/GB200/GB300 quoted "Contact us"; all reserved dedicated capacity is "Contact sales"). On-demand GPU clusters: HGX H100 at $3.99/hr, HGX H200 at $5.99/hr, HGX B200 at $8.19/hr. Reserved clusters (7–30 day commits): H100 at $3.59/hr, H200 at $4.99/hr, B200 at $7.99/hr, dropping to $3.09/hr H100 on 91–180 day reservations — the reserved rates are among the lowest published in managed inference. Separately, a July 2026 Provisioned Throughput (PTU) SKU reserves capacity in throughput units at $0.05 per PTU-minute (MiniMax M3, GLM-5.2) for buyers who prefer a fixed tokens-per-minute envelope over per-GPU-hour rentals.
Does Together AI have a free tier? ▼
New accounts can start without an upfront commitment, but Together does not publish a specific signup-credit dollar amount on its pricing page or quickstart docs. Calling paid serverless and image models requires a positive credit balance, and production usage requires a payment method on file. There is no permanent free tier.
Is this Together AI pricing calculator free and accurate? ▼
Yes — it's completely free with no signup. It uses the latest publicly available Together AI pricing (verified July 2026). Actual costs may vary with volume discounts or enterprise terms — always confirm on Together AI's official pricing page.