Compare
Side by side.
Pick up to three companies to compare current plans and recent moves.
Together AI
Together AI charges based on usage for various services like serverless inference, dedicated inference, GPU clusters, sandbox compute, and fine-tuning, with prices varying by model, hardware, and duration.
Serverless Inference (Kimi K3)$3.00per 1M input tokens
- Output: $15.00 per 1M tokens
- Cached output: $0.30 per 1M tokens
Serverless Inference (Ternary Bonsai 27B)$0.00per 1M input tokens
Dedicated Inference (NVIDIA HGX H100)$5.49per gpu per hour
- Guaranteed performance
- Support for custom models
- Autoscaling & traffic spike handling
GPU Clusters (NVIDIA HGX H100)$3.99per GPU per hour
- On-demand pay as you go
Sandbox (Per vCPU)$0.0446per hour
- Customize VM sandboxes
Post-Training (NVIDIA HGX B200)$9.00per GPU per hour
- On-demand training infrastructure
- Promotion valid until 12/14/2026
Recent changes
P0 Together AI Updates Fine-Tuning and Inference Pricing, Introduces New Models and Post-Training Service
DeepSeek API
DeepSeek API charges based on the number of input and output tokens for models, and per 1,000 successful requests for Search and Fetch APIs, with different rates for peak and off-peak hours.
deepseek-flash (1M INPUT TOKENS - CACHE HIT)$0.003per 1M tokens
- Off-peak rate
- Peak rate: $0.006
deepseek-flash (1M INPUT TOKENS - CACHE MISS)$0.15per 1M tokens
- Off-peak rate
- Peak rate: $0.3
deepseek-flash (1M OUTPUT TOKENS)$0.6per 1M tokens
- Off-peak rate
- Peak rate: $1.2
deepseek-v4-pro (1M INPUT TOKENS - CACHE HIT)$0.022per 1M tokens
- Off-peak rate
- Peak rate: $0.044
deepseek-v4-pro (1M INPUT TOKENS - CACHE MISS)$0.66per 1M tokens
- Off-peak rate
- Peak rate: $1.32
deepseek-v4-pro (1M OUTPUT TOKENS)$1.98per 1M tokens
- Off-peak rate
- Peak rate: $3.96
Search API$0.75Per 1,000 successful requests
Fetch API$0.75Per 1,000 successful requests
Recent changes
P0 DeepSeek API introduces new Search / Fetch API and detailed model pricing tiers