Together AI

What is Together AI?

Together AI is a cloud platform for developers who build with open-source models. Serverless inference lets you call more than 100 models through one API with no minimum spend and no provisioning, and you pay per token for chat, embedding and rerank models, per megapixel or image for image models, per second for video, and per second of audio for speech. Batch inference runs asynchronous jobs at up to 50% lower cost on discounted models. When you need guaranteed capacity, provisioned throughput reserves throughput units, and dedicated inference gives you single-tenant GPU endpoints with autoscaling and support for custom models. The products page also lists GPU clusters, code sandboxes, managed storage with zero egress fees, and fine-tuning. Serverless pricing is published per model, for example gpt-oss-120B at $0.15 input and $0.60 output per 1M tokens. Scale and Enterprise plans add custom rate limits, HIPAA support and availability SLAs.

Use cases

• Call open-source LLMs through one API
• Run large batch inference jobs at lower cost
• Host models on dedicated GPU endpoints
• Fine-tune open models on your own data
• Generate images, video and speech via API
• Run code in sandboxes for AI agents
• Rent GPU clusters for training

Who it's for

• Developers
• AI engineers
• Product teams
• Enterprises

Why use Together AI

• Pay-per-use serverless means no minimum cost and no hardware to manage while you prototype.
• The same inference APIs work on serverless and dedicated hardware, so you can scale up without rewriting code.
• Batch pricing cuts the cost of workloads that do not need real-time responses.
• One vendor covers inference, GPU compute and fine-tuning for open models.

Together AI features

• Serverless inference across 100+ open-source models
• One API for chat, vision, image, audio, video, embeddings, rerank and moderation
• Batch inference at up to 50% lower cost for eligible models
• Provisioned throughput billed in throughput units
• Dedicated model inference on single-tenant GPUs with autoscaling
• Dedicated container inference for image, audio and video models
• GPU clusters, including on-demand B200s
• Code sandboxes and managed storage with zero egress fees
• Fine-tuning and post-training for open models
• Geo-specific inference in the United States or Canada

How to get started with Together AI

1. Pick a model from the serverless catalog and make a first API request with your key.
2. Pay per token, megapixel, image, second or audio minute depending on the model type.
3. Send non-urgent workloads through the batch API for discounted processing.
4. Move to provisioned throughput or a dedicated endpoint when you need reserved capacity.
5. Use fine-tuning or GPU clusters if you need to adapt or train a model.

Together AI limitations

• Serverless requests can hit 429 or 503 responses when demand is high
• Geo-specific inference costs 1.25x the standard serverless rate
• Image prices vary by provider and settings, and listed prices use the lowest resolution or duration

Together AI pricing

Together AI has no free plan.

Together AI plans

• Serverless Inference

  Pay per use

  Per token, megapixel, second or image by model

  • No minimum cost
  • 100+ open-source models
  • Batch API discounts up to 50%

• Provisioned Throughput

  Price on vendor site

  • Reserved throughput units
  • Uptime SLA

• Dedicated Inference

  Price on vendor site

  • Single-tenant GPU instances
  • Autoscaling
  • Custom model support

• Scale

  Custom

  • Custom rate limits
  • Discounts
  • HIPAA
  • 99% availability SLA

• Enterprise

  Custom

  • All Scale benefits
  • 99.9% SLA
  • Custom regions
  • Priority hardware access

Plans as listed on the official pricing page on 24 Oct 2026. Prices can change and may vary by country or exclude tax. Check the official pricing page.

Together AI alternatives

The main Together AI alternatives listed on Debuly are Fireworks AI, Replicate, Groq, Anyscale, OpenRouter and Baseten. The table compares pricing and platforms; order is A–Z.

1. 

   Anyscale

   Managed platform for running Ray workloads, with pay-as-you-go compute and options to deploy in your own cloud.

   Developer platformsAutomation and agents
   PaidVisit Anyscale (official site)
2. 

   Baseten

   Inference platform for deploying open-source, custom and fine-tuned AI models, with per-token Model APIs and dedicated GPUs.

   Developer platforms
   PaidVisit Baseten (official site)
3. 

   Fireworks AI

   Inference and training platform for open models: pay-per-token serverless APIs, GPU deployments, fine-tuning and reserved capacity.

   Developer platformsCodingAutomation and agents
   PaidVisit Fireworks AI (official site)
4. 

   Groq

   GroqCloud inference API that serves open models such as GPT-OSS and Whisper on Groq's LPU hardware, billed per token.

   Developer platformsAudio and voiceCoding
   FreemiumVisit Groq (official site)
5. 

   OpenRouter

   Unified OpenAI-compatible API for 500+ AI models with provider fallbacks, pay-as-you-go credits and a free tier

   Developer platformsChat assistantsAutomation and agents
   FreemiumVisit OpenRouter (official site)
6. 

   Replicate

   Cloud API for running, fine-tuning and deploying open-source and proprietary AI models, billed by compute time or by output.

   Developer platformsImage generationVideo
   PaidVisit Replicate (official site)

Together AI FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from cdn.prod.website-files.com · fetched 24 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.