Deploy and scale on
the AI cloud
Inference by the token. Compute by the second. Scale by the month — or don't. We run the infra so you can ship the model.
Trusted by teams at
Inference API
Inference Engine
200+ models across chat, code, image, video, audio, and more. One endpoint, every provider.
Chat & Reasoning
80+ models
Code Generation
25+ models
Image Generation
30+ models
Video & Motion
15+ models
Audio & Speech
20+ models
Vision & OCR
15+ models
Embeddings
12+ models
Transcription
8+ models
Infrastructure
Compute
Raw bare metal. Full root access. Pick your hardware, deploy in 60 seconds. Per-minute billing. Scale from 1 node to 128.
Available Hardware
Current fleet
Performance
Key specs
Root Access
Full control over your environment. SSH, Docker, custom images.
Auto-scaling
Scale horizontally on demand. Add nodes in seconds, release when done.
Pay Per Minute
No minimum commitments. Spin up for 5 minutes or 5 months.
Platform
The last AI cloud you'll sign up for.
(Not another provider to reconcile.)
Inference and compute. One credit balance.
Add funds once. Spend on API calls or GPU hours — same balance, same dashboard. No reconciling invoices from three providers.
Self-Serve
Get an API key. Ship today. Talk to nobody.
$0
per month · pay as you go
What's included
- Pay-per-token inference on every model
- Per-second GPU compute
- OpenAI-compatible endpoint
- Public rate card, no negotiations
Deployment
Dedicated
Most popularWhen the rate card starts to hurt.
Custom
volume discounts
What's included
- Everything in Self-Serve
- 40–60% off the public rate card
- Reserved GPU capacity
- 99.9% uptime SLA
Deployment
Enterprise
Run it our way. Or run it in your VPC.
Custom
tailored contract
What's included
- Everything in Dedicated
- BYOC + self-hosted deployments
- Region pinning — US / EU / APAC
- Named CSM + on-call engineering
Deployment
From frontier training to production inference.
One platform for every AI workflow. Your team's default.
ManagedPre-training at scale.
Multi-node clusters with NVLink fabric. Submit jobs via squeue. We handle the orchestration.
One platform for training, finetuning, and inference.
Whether you're pre-training a frontier model, finetuning on proprietary data, or serving production traffic — it all runs on the same infrastructure. One dashboard. One bill.
Managed Finetuning
Upload dataset, pick base model. Pay per training second.
Dedicated Inference
Reserved GPUs, p99 SLAs. 40–60% cheaper than hyperscalers.
Managed Slurm
Multi-node clusters. Hundreds of GPUs, one command.
api.runcrate.ai/v1/chat/completionsWhy Runcrate
One platform. Every AI workload.
Inference API and GPU compute, unified. One API key, one bill, one dashboard. Stop juggling providers.
200+ models, one endpoint.
DeepSeek, Llama, Claude, Qwen, FLUX, Sora — all via OpenAI-compatible API. Swap base_url and ship.
GPUs in 60 seconds.
H100, H200, B200, MI300X — per-second billing, no commitments. Stop the instance, stop the meter.
40–60% cheaper.
Public rate card beats every aggregator. Volume discounts on dedicated. No hidden fees, credits never expire.