Criterion GroqCloud Replicate
CategoryInference & Model HostingAI Deployment & Monitoring
In one sentenceLow-latency inference API for open models running on Groq's LPU hardware.Run and deploy open-source and custom AI models through an API, paying per second of hardware or per output.
Problem it solvesGetting fast responses from open models for real-time apps without hosting them.Using or deploying AI models, especially image and video models, without managing GPUs.
Free planYesNo
Developer APIYes — docsYes — docs
Open sourceNoNo
PlatformsWeb (Groq console), REST APIWeb, REST API
Pricing (summary)Free plan with rate limits, then usage-based pricing per 1M tokens on the Developer plan. For example, openai/gpt-oss-120b is $0.15 input / $0.60 output and gpt-oss-20b $0.075 / $0.30; some models are contact-sales.
Checked 2026-10-19		Usage-based. Public models mostly bill by hardware time (e.g. Nvidia T4 $0.81/hr, A100 80GB $5.04/hr, H100 $5.49/hr, billed per second). Some official models bill per output or token, such as flux-1.1-pro at $0.04 per image.
Checked 2026-10-19
Strengths
  • Speed-focused hardware
  • Free entry point
  • Simple per-token prices for gpt-oss models
  • Simple pay-per-use pricing
  • Huge model selection
  • Good for media models
Limitations
  • Several models list pricing as 'Contact Sales'
  • Free-plan limits not published in docs
  • Preview models may be discontinued at short notice
  • No free tier mentioned on the pricing page
  • Private deployments bill idle time
  • Costs vary by model runtime and can be hard to predict