Together AI pricing & changelog
Sign in to trackInference, fine-tuning, and model hosting for AI teams.
At a Glance: Together AI Pricing in 2026
Starting Price
Custom / Usage
Free Tier
Paid Plans Only
Total Tiers Tracked
7 plan(s)
How much does Together AI cost in 2026?
Verified Source: Extracted directly from Together AI Official Pricing Page. Verified and tracked daily for changes by TrackRival.
Serverless Inference
—
- • Pay-as-you-go model APIs
- • Input and output pricing per 1M tokens
- • Wide variety of chat, vision, and code models
- • Caching discounts available on select models
Provisioned Throughput
—
- • Reserve dedicated capacity in throughput units (PTUs)
- • Fixed capacity with predictable performance
- • Continuous 24/7 provisioning options
- • Significant savings compared to commercial list prices
Dedicated Inference
Custom
- • Single-tenant GPU instances with guaranteed performance
- • No resource sharing
- • Support for custom models
- • Autoscaling & traffic spike handling
GPU Clusters
—
- • On-demand hourly GPU capacity
- • Preemptible compute options available
- • Reserved capacity discounts for 3 to 181+ days
- • Pricing starting from $1.99 per GPU per hour
Sandbox & Code Interpreter
—
- • VM sandboxes for large development environments
- • Priced at $0.0446 per vCPU per hour
- • Priced at $0.0149 per GiB RAM per hour
- • Secure Code Interpreter sessions at $0.03 for 60 minutes
Managed Storage
—
- • High-bandwidth, parallel filesystem
- • Colocated with your compute
- • Priced at $0.16 per GiB/month
Post-Training & Fine-Tuning
—
- • On-demand training infrastructure billed per minute
- • Supervised Fine-Tuning (SFT) and Direct Preference Optimization
- • Priced per 1M tokens based on training dataset
- • Minimum charges apply per job depending on the model
What recent features and updates did Together AI ship?
No changelog entries tracked yet.
What are the best alternatives to Together AI?
Frequently Asked Questions about Together AI Pricing
- How much does Together AI cost in 2026?
- Together AI provides 7 plan tiers (Serverless Inference, Provisioned Throughput, Dedicated Inference, GPU Clusters, Sandbox & Code Interpreter, Managed Storage, Post-Training & Fine-Tuning), with starting pricing at Custom / Usage. Detailed feature breakdowns and billing periods are updated daily above.
- Does Together AI have a free tier or trial?
- No, Together AI does not advertise a free plan; starting prices begin at Custom / Usage.
- What are the hidden costs or usage overage fees for Together AI?
- Together AI's total cost often depends on usage parameters such as active users, API request volume, compute instance hours, or storage consumption. Review the latest plan details above before choosing a tier.
- What are the top alternatives to Together AI?
- Leading competitor alternatives in the ai space include OpenAI, Anthropic, Replicate, OpenRouter, Mistral. Compare them side-by-side on TrackRival.
- How often is Together AI's pricing verified?
- TrackRival monitors Together AI's official pricing page daily via automated Playwright scrapers and logs any changes to plans, quotas, and release notes immediately.
Embed Together AI Verification Badge
Showcase that Together AI's pricing and releases are monitored by TrackRival on your GitHub README, documentation, or website.
Markdown (GitHub README / Docs):
[](https://trackrival.dev/product/together-ai)
HTML (Website Footer / Docs):
<a href="https://trackrival.dev/product/together-ai"><img src="https://trackrival.dev/api/badge/together-ai" alt="Together AI tracked on TrackRival" /></a>