# Nodus blog

> Guides, prices and benchmarks for AI training, inference and coding agents. Each link is the Markdown version of a post.

## Posts

- [Nodus vs E2B: which sandbox should AI agents use?](https://www.nodus-compute.ai/docs/source/blog/nodus-vs-e2b.md) (2026-10-14): Nodus vs E2B for AI agent sandboxes: isolation, what the defaults do when an agent forgets, network access, GPUs and what 2,000 sandbox-hours cost.
- [How much VRAM do you need to run or fine-tune a 70B model?](https://www.nodus-compute.ai/docs/source/blog/vram-needed-for-70b-model.md) (2026-10-13): A 70B model needs 40 to 141 GB for weights plus a KV cache to serve, and 50 GB to 1.1 TB to fine-tune. The math, GPU setups and a script to check yours.
- [Nodus vs Modal: which is better for GPU training?](https://www.nodus-compute.ai/docs/source/blog/nodus-vs-modal.md) (2026-10-12): Nodus vs Modal for GPU training: listed rates, preemption, the 24-hour limit, multi-node clusters, agent support and what the same jobs cost on each.
- [How do you stop an AI agent from overspending on GPUs?](https://www.nodus-compute.ai/docs/source/blog/ai-agent-gpu-spending-limits.md) (2026-10-11): A prompt is not a limit. Cap every run, put a blocking budget around agent work, scope its credentials and approve writes. Setup for Claude Code and Codex.
- [Continued pretraining on Nodus: 42.5% lower charges](https://www.nodus-compute.ai/docs/source/blog/nodus-vs-modal-continued-pretraining-benchmark.md) (2026-10-11): Qwen2.5-0.5B on one L4: 42.5% lower recorded execution charges in our continued-pretraining pilot. Inspect the evidence and offline cost worksheet.
- [Fine-tuning on Nodus: 67% lower charges](https://www.nodus-compute.ai/docs/source/blog/nodus-vs-modal-fine-tuning-benchmark.md) (2026-10-11): Qwen2.5-0.5B on one L4: 67% lower successful-run charges, 24.7% including a failed Nodus attempt. Inspect the pilot records and offline cost worksheet.
- [Spot vs on-demand GPUs: what breaks, and how do you make spot safe?](https://www.nodus-compute.ai/docs/source/blog/spot-vs-on-demand-gpus.md) (2026-10-10): Spot GPUs cost up to 90% less but can vanish with 30 seconds of notice. What breaks, each cloud's notice window, the cost math and a safe setup.
- [What are the best Modal alternatives for GPU training in 2026?](https://www.nodus-compute.ai/docs/source/blog/modal-alternatives-for-training-2026.md) (2026-10-09): Modal alternatives for training compared on listed H100 prices, preemption handling, run limits and cost per run, as of October 9, 2026.
- [Cloud GPU pricing, October 2026: what an H100, H200, B200 or B300 costs per hour](https://www.nodus-compute.ai/docs/source/blog/cloud-gpu-pricing-october-2026.md) (2026-10-08): Listed per GPU-hour prices for B300, B200, H200, H100, A100, L40S, L4 and RTX cards as of October 8, 2026, with a training-run cost example.
- [How to give Claude Code, Codex or Cursor access to GPUs (MCP setup)](https://www.nodus-compute.ai/docs/source/blog/give-claude-code-codex-cursor-gpus-mcp.md) (2026-10-08): Connect one MCP server and your coding agent can estimate, launch and watch GPU jobs on your account, with dry runs and spending limits.
- [How to make GPU training survive spot preemption (without babysitting it)](https://www.nodus-compute.ai/docs/source/blog/resume-training-after-spot-preemption.md) (2026-10-08): Save model, optimizer and step to one directory, write atomically, resume on start, and let the platform carry that directory across a reclaim.

## Connect a coding agent

- [Setup for MCP clients](https://www.nodus-compute.ai/connect.md)