Submit your tool

Built for devs.

Twitter Submit a tool Share feedback

Built for devs.

Twitter•Submit a tool•Feedback

Together AI

Fast inference for 200+ open source models with fine-tuning and GPU clusters

Usage-based LLM APIsLast verified: 2026-01-25

Overview

Together AI is a cloud platform for generative AI providing fast serverless inference, fine-tuning, and GPU clusters. Access 200+ open source models including Llama, DeepSeek, Qwen, Mistral, and FLUX through OpenAI-compatible APIs. Serverless inference is usage-based with per-token pricing.

Works with

REST APIOpenAI compatibleLangChainLlamaIndex

Pricing

MOST POPULAR

Usage-basedServerless Inference

200+ models available
Pay per token
OpenAI-compatible API
Auto-scaling
Llama 3.1 8B: $0.18/1M tokens
Llama 3.3 70B: $0.88/1M tokens
DeepSeek-R1: $3.00 in / $7.00 out
Batch API at 50% discount

From $2.10/hrDedicated Endpoints

Guaranteed performance
Custom models supported
Autoscaling
L40S 48GB: $2.10/hr
A100 80GB: $2.40-2.56/hr
H100 80GB: $3.36/hr
H200 141GB: $4.99/hr
99.9% SLA available

From $0.48/1M tokensFine-tuning

LoRA and full fine-tuning
Up to 16B: $0.48/1M (LoRA)
17B-69B: $1.50/1M (LoRA)
70B-100B: $2.90/1M (LoRA)
DPO training supported
DeepSeek models: $10/1M tokens

From $1.75/hr/GPUGPU Clusters

H100 SXM: $2.20-2.99/hr
H200: $3.15-3.79/hr
B200: $4.50-5.50/hr
Kubernetes or Slurm
Free network ingress/egress
NVLink networking

Pros

+200+ open source models including Llama DeepSeek Qwen Mistral
+4x faster than vLLM with Together Inference Engine
+OpenAI-compatible API for easy migration
+Fine-tuning with LoRA and full training
+Batch API at 50% discount
+Image and video generation included
+GPU clusters with H100 H200 B200 hardware

Cons

-No free tier or credits (pure usage-based)
-Pricing varies widely by model size
-Large models (405B DeepSeek-R1) can get expensive
-Dedicated endpoints require minimum commitment

Find similar tools

Anthropic

Claude 4.5 family - Opus Sonnet Haiku

Anthropic

Claude 4.5 family - Opus Sonnet Haiku

Cerebras

Ultra-fast LLM inference powered by wafer-scale chips

REST APIOpenAI compatible

Cerebras

Ultra-fast LLM inference powered by wafer-scale chips

Get Discovered by Developers

Promote your tool

Reach thousands of developers actively searching for AI tools. Featured listings get 10x more clicks.