AI Tool
Together AI pricing, features, company info, and alternatives
A factual product page for Together AI as an AI infrastructure platform for open-source models.
Last updated April 2026 · Pricing and features verified against official documentation
Pricing
Current public pricing tiers on file for Together AI, last verified Apr 21, 2026.
Serverless Inference
Usage-based
Model pricing varies by model and modality; the pricing page lists per-1M-token rates and a $5 minimum credit purchase applies to platform access.
Dedicated Model Inference
From $3.99/hour
The pricing page lists 1x H100 80GB at $3.99/hour, with H200 and B200 options also available.
GPU Clusters
From $3.49/hour
On-demand HGX H100 pricing starts at $3.49/hour; H200 and B200 options are also listed.
Fine-Tuning
From $0.48/1M tokens
Supervised fine-tuning LoRA for models up to 16B starts at $0.48 per 1M tokens.
Sandbox
From $0.0446/vCPU-hour
The code sandbox page also lists $0.0149/GiB RAM-hour and a $0.03/session code interpreter.
What You Can Do With It
The main capabilities that shape how people use Together AI today.
Runs open-source models through an OpenAI-compatible API across chat, image, vision, audio, embeddings, rerank, and moderation surfaces.
Offers separate serverless and dedicated inference paths for usage-based workloads and reserved infrastructure.
Includes fine-tuning, dedicated container inference, GPU clusters, and managed storage in the same platform family.
Documents sandboxed code execution and model-building workflows for AI application development.
Best For
Who Together AI is most clearly built for.
Teams shipping production AI apps that need one vendor for inference, compute, and model shaping.
Developers who want to move from shared serverless usage to dedicated infrastructure on the same platform.
Organizations that need to fine-tune open-source models and run custom workloads without managing underlying GPUs.
Company
Leadership and company context for Together AI.
CEO
Vipul Ved Prakash
Founders
Vipul Ved Prakash, Ce Zhang, Chris Ré, Tri Dao, Percy Liang
Investors
General Catalyst, Prosperity7, Salesforce Ventures, NVIDIA, Kleiner Perkins
Platforms
Where you can use Together AI today.
Web
API
Python SDK
TypeScript SDK
Privacy Notes
Publicly stated data-handling notes that matter when evaluating Together AI.
The privacy policy says users can enable Zero Data Retention by choosing not to store prompts or allow data to train models.
Under Zero Data Retention, submitted content and outputs are not stored, retained, or used for model training or product improvements.
The policy says usage data may still be retained for internal analysis and to improve security or functionality.
Compliance
Public compliance or enterprise-governance signals we found for Together AI.
SOC 2 Type II
HIPAA
GDPR
Access
How to integrate or build around Together AI.
Public API
Yes
Docs
Available
Alternatives
Other tools worth considering alongside Together AI.
Unified API and chat layer for routing across hundreds of AI models and providers.
AWS's managed foundation-model platform for building, customizing, and governing generative AI apps.
Browser-based Gemini workspace for prototyping, testing, and shipping AI apps.
European AI platform with Le Chat, model APIs, and enterprise AI Studio tooling.
Product Snapshot
Together AI is an AI infrastructure platform for running, fine-tuning, and training open-source models. It combines serverless inference, dedicated inference, GPU clusters, sandboxed code execution, storage, and fine-tuning in one product family.
What You Can Do With It
- Run models through an OpenAI-compatible API across chat, image, vision, audio, embeddings, rerank, and moderation surfaces.
- Move from serverless usage to dedicated endpoints and isolated GPU clusters for steadier workloads.
- Fine-tune open-source models and run custom workloads on Together-managed infrastructure.
- Use code sandboxes and managed storage for app development, evaluation, and training jobs.
Why It Stands Out
Together AI packages inference, compute, and model shaping into a single cloud with separate paths for usage-based and reserved infrastructure.
Tradeoffs To Know
- The public pricing surface mixes model-specific token rates with infrastructure pricing, so cost planning requires reading the exact service you plan to use.
- Serverless model pricing varies by model and modality, so there is no single flat API rate.
- Enterprise deployment and support options are described across multiple product and docs pages, so buyers should verify the exact plan before committing.
Changes to this tool page
- April 2026 Initial page created from Together AI's official product, pricing, docs, privacy, and company pages.
Sources
- together.ai/pricing
- docs.together.ai/docs/billing
- docs.together.ai/docs/fine-tuning-pricing
- together.ai/dedicated-model-inference
- together.ai/serverless-inference
- together.ai/about-us
- together.ai/privacy
- together.ai/blog/together-ai-announcing-305m-series-b
- together.ai/blog/api-announcement
- together.ai/blog/series-a
- together.ai
- docs.together.ai/docs/introduction
- together.ai/blog/introducing-the-together-enterprise-platform
- together.ai/openai