Alternatives to RunPod

Modal's serverless GPU cloud runs Python functions and sandboxes with per-second billing, spinning up GPU inference and agent compute from

Run image, video, speech and LLM models through one Replicate API and one bill

Lilac turns idle GPU clusters into a white-labeled neocloud with spot H100 leases and serverless inference.

Salad Cloud rents idle consumer GPUs from $0.015 per GPU-hour, billed per second, for retryable inference and batch work.

Together AI is an AI cloud for open-weight models: per-token inference APIs with GPU clusters underneath.

Serverless GPU inference for custom ML models — deploy in minutes, billed per second with scale-to-zero.

Crusoe Cloud GPU instances, serverless inference, and fine-tuning priced for AI teams that want orchestration included

Serverless inference API for 100+ image, video and 3D models — FLUX, Veo 3.1, Wan 2.1 — billed per output or per GPU-second.

One inference API for 400K+ image, video, audio, LLM and 3D models, plus serverless GPUs starting at $0.63/GPU-hour reserved.

Serverless AI inference on 100+ open models via an OpenAI-compatible API, billed per token with no GPU overhead.

Cerebras inference cloud runs frontier LLMs in Ultrafast mode at up to 750 tokens/sec for real-time agents and voice AI.

Rent GPUs on Vast.ai's marketplace — 20,000+ cards, per-second billing, and API-native provisioning agents can drive.

CoreWeave is the AI-native GPU cloud for training, reinforcement learning post-training, and low-latency inference on NVIDIA cluster-scale

Run untrusted code and AI workloads on serverless GPU compute with forkable microVMs and millisecond-billed sandboxes.

Modular's unified inference stack delivers hardware-portable LLM serving from GPU kernel to OpenAI-compatible cloud endpoint

Multimodal inference API on dedicated NVIDIA GPUs with an OpenAI-compatible endpoint

ViewComfy turns ComfyUI workflows into no-code web apps and autoscaling serverless APIs, billed per active GPU second.

vLLM turns open-weight LLMs into high-throughput, OpenAI-compatible inference endpoints on your own GPUs.

Run 20+ pre-installed open-source AI apps — ComfyUI, Stable Diffusion, RVC, LoRA training — in your browser on rented cloud GPUs

European GPU cloud with self-service GB300, B300 and B200 instances plus per-token open-weight inference

EU-sovereign GPU cloud for AI training and inference — Hyperstack on-demand or single-tenant Secure Private Cloud.

Run Llama 3.1 405B and other huge language models at home over a BitTorrent-style decentralized inference network.

Sovereign AI cloud: Nscale owns the data centers and power, then sells reserved GPU capacity, managed inference and fleet operations on top.

Rent GPU instances and per-second GPU sandboxes for AI training, RL rollouts, and agent workloads from $0.35/hr.

Per-second-billed cloud GPUs (H100, A100, A6000, V100) with 1-click Jupyter notebooks, model deployment, and distributed training in beta.

Krutrim Cloud rents A100 and H100 GPU instances in India with rupee billing and Indian data residency.

Etched builds custom inference silicon and rack-scale frontier inference clusters for trillion-parameter MoE prefill and decode.

Enterprise LLM fine-tuning with accuracy SLAs and sub-second inference.

Rack-scale AI inference system using log-space math to serve multi-trillion-parameter models at 608 PFLOPS per rack.

Groq delivers fast LLM inference on its own LPU and LPX silicon through an OpenAI-compatible API.