# Arrakis vs Modal Modal runs Python functions and web endpoints on GPUs that scale to zero, billed by the second. Arrakis rents whole GPU machines with root SSH at the lowest live price across independent markets, billed by the hour from a prepaid balance. ## Side by side As of 31 Oct 2026, 17:40 UTC, the lowest price for the NVIDIA B300 on Arrakis is $4.60 per GPU-hour, with 73 GPUs available on 10 machines. | | Arrakis | Modal | |---|---|---| | B300 per GPU-hour | Live: [GPU prices](/gpus.html.md) | $7.10 ($0.001972 a second) | | B200 per GPU-hour | Live: [GPU prices](/gpus.html.md) | $6.25 ($0.001736 a second) | | H100 per GPU-hour | Live: [GPU prices](/gpus.html.md) | $3.95 ($0.001097 a second) | | Billing | By the hour; a started hour counts in full | Per second, scaling to zero | | How you pay | Prepaid balance: card, Link or Apple Pay | Card charged monthly | | Root SSH and any Docker image | Yes | No SSH by default: modal shell | | Serverless endpoints, billed per request | No: each model gets a GPU of its own | Yes: functions and web endpoints | | Hosted models, billed per output | No: deploy any Hugging Face model yourself | Language models only, per token | | MCP server for AI agents | Yes, remote at /mcp | No official server: agent skills | | Spend cap on each API key | Yes, on every token | Budgets per workspace, not per key | | Automatic shutdown | Yes: every rental has a removal time | Yes: idle containers scale down | | Interruptible machines | Yes, where hosts offer them | No discount tier | | Searches several GPU markets | Yes | No: its own platform | Facts about other providers come from their own pages, checked on October 31, 2026. Prices and features change: check them before you choose. Arrakis is not affiliated with them. ## Which fits you - **Modal fits better.** Your work is bursty Python code that should scale to zero between requests, and you'd rather not manage machines at all. - **Arrakis fits better.** You want a whole machine with root SSH for hours or weeks, the lowest live hourly price, any Docker image as it is, or an agent that rents within a hard cap. ## Moving to Arrakis Your Docker image, SSH keys and scripts move as they are: Arrakis boots any public or private image with root SSH, and serves your own model with your own engine options. Start with one machine in the console, or point your agent at /mcp with a capped token. ## Questions **Is Modal or Arrakis cheaper for GPUs?** For steady work, a machine by the hour can cost less: on October 31, 2026, Modal listed an H100 at $0.001097 a second, about $3.95 an hour, billed only while code runs. For bursty traffic that sits idle most of the day, paying by the second can cost less. **Can I SSH into a Modal container?** Not by default: Modal offers modal shell to open a shell in a container. Arrakis gives root SSH into every machine. ## Sources - - - - - -