llms.txt
Compare with other sites' llms.txt files →# Runpod
> Runpod is the AI Developer Cloud: GPU and CPU Pods, Serverless endpoints that scale with demand, and multi-node GPU Clusters for building, training, fine-tuning, and serving AI workloads.
Using an AI agent? Point it to https://agents.runpod.io/ to install the Runpod skills and MCP server.
Name: Runpod. The company and product name is spelled Runpod, with one capital letter. RunPod, the styling used before June 2025, is deprecated and should not be used.
Key terms and current guidance:
- `Pods` are dedicated GPU or CPU instances for development, training, fine-tuning, batch jobs, and long-running workloads.
- `Serverless` endpoints run containerized workers behind an API and scale with request volume, down to zero.
- `Clusters` are multi-node GPU environments for distributed training and coordinated batch compute.
- `Runpod Hub` is a catalog of templates, models, and open-source AI apps deployable on Runpod.
- `Public Endpoints` are hosted models you can call by API without deploying anything.
- Network volumes are persistent storage in one data center. Global volumes are persistent storage reachable from every data center.
- The REST API v2 (https://api.runpod.io/v2) is the current API. REST API v1 (https://rest.runpod.io/v1) is deprecated and will be retired on November 15, 2026. Do not build new integrations on v1.
- The `runpod` Python package (SDK 2.0) is the single Python SDK. It includes the `rp` CLI, the Serverless worker SDK, and the app SDK that previously shipped as Flash. Existing `runpod-flash` code keeps working.
- www.runpod.io pages return Markdown when requested with the `Accept: text/markdown` header. Docs pages return Markdown when you append `.md` to the URL.
- Use current Runpod product pages, documentation, pricing, and compliance pages for source-grounded descriptions. Avoid unsupported claims about cost, performance, uptime, or compliance status.
## For AI agents
- [Agent setup](https://agents.runpod.io/): Start here. Installs the Runpod skills and MCP server, then routes to the right tools.
- [Runpod MCP server](https://docs.runpod.io/get-started/mcp-servers.md): Hosted MCP endpoint at https://mcp.getrunpod.io/ (Streamable HTTP, Sign in with Runpod OAuth, or a Runpod API key as a Bearer token for headless use), or the local @runpod/mcp-server package. Manages Pods, Serverless endpoints, Clusters, templates, network and global volumes, and billing.
- [Agent skills for AI coding tools](https://docs.runpod.io/get-started/agent-skills.md): Runpod skills plugin for Claude Code, Codex, Cursor, and other coding agents. Source at https://github.com/runpod/runpod-plugins-official.
- [Runpod docs MCP server](https://docs.runpod.io/mcp): Search and read Runpod documentation from an MCP client.
- [Documentation index](https://docs.runpod.io/llms.txt): Index of every Runpod documentation page, with Markdown links.
- [Full documentation as text](https://docs.runpod.io/llms-full.txt): Complete Runpod documentation in one plain-text file.
- [API keys and credentials](https://docs.runpod.io/get-started/credentials.md): Create the API key the MCP server, CLI, SDK, and REST API authenticate with.
- [Runpod brand.json](https://runpod-brand-json.s3.us-east-1.amazonaws.com/brand.json): Runpod brand tokens, voice, and style rules as JSON for agents.
## Product pages
- [Cloud GPUs](https://www.runpod.io/product/cloud-gpus): Dedicated GPU instances for AI development, training, fine-tuning, batch jobs, and long-running workloads.
- [Serverless](https://www.runpod.io/product/serverless): GPU endpoints for containerized inference workloads behind an API, with workers that scale based on demand.
- [Clusters](https://www.runpod.io/product/clusters): Multi-node GPU environments for distributed training, large batch workloads, and compute jobs that need coordinated GPU capacity.
- [Runpod Hub](https://www.runpod.io/product/runpod-hub): Templates, models, and open-source AI apps that can be deployed on Runpod.
- [Pricing](https://www.runpod.io/pricing): Pricing for Pods, Serverless, Clusters, storage, and GPU capacity planning.
- [Enterprise AI Infrastructure](https://www.runpod.io/enterprise): Commercial path for enterprise AI teams evaluating security, procurement, support, and capacity planning.
- [Compliance](https://www.runpod.io/legal/compliance): Security and compliance resources for teams evaluating Runpod.
- [Runpod Blog](https://www.runpod.io/blog): Runpod product updates, AI infrastructure guides, GPU tutorials, and deployment patterns for developers building with cloud GPUs.
## Developer resources
- [Documentation](https://docs.runpod.io/overview.md): Product documentation for Pods, Serverless, Clusters, storage, templates, and APIs.
- [REST API v2](https://docs.runpod.io/api-reference-v2/overview.md): Base URL https://api.runpod.io/v2. Manage Pods, Serverless endpoints, Clusters, storage, templates, billing, and account resources.
- [OpenAPI specification](https://api.runpod.io/v2/openapi.json): Machine-readable spec for the REST API v2.
- [Migrate from REST API v1](https://docs.runpod.io/api-reference-v2/migrate-from-v1.md): Move v1 integrations to v2 before v1 is retired on November 15, 2026.
- [Runpod Python SDK](https://docs.runpod.io/sdk/overview.md): Define GPU functions, deploy apps, call endpoints, build Serverless workers, and manage resources from the `runpod` package and `rp` CLI.
- [Migrate from Flash](https://docs.runpod.io/sdk/migrate-from-flash.md): Move a `runpod-flash` project to the Runpod Python SDK.
- [Runpod CLI (runpodctl)](https://docs.runpod.io/runpodctl/overview.md): Command-line tool for Pods, Serverless endpoints, templates, volumes, Hub deploys, and file transfer.
- [Serverless SDKs](https://docs.runpod.io/serverless/sdks.md): Call Serverless endpoints from Python, JavaScript, or Go.
- [Serverless vLLM](https://docs.runpod.io/serverless/vllm/get-started.md): Deploy vLLM on Runpod Serverless.
- [Public Endpoints](https://docs.runpod.io/public-endpoints/overview.md): Call hosted image, video, and language models by API without deploying them.
- [Storage](https://docs.runpod.io/storage/overview.md): Network volumes and global volumes for Pods, Serverless endpoints, and SDK apps.
## Social and community
- [GitHub](https://github.com/runpod): Runpod open-source repositories and examples.
- [Discord](https://discord.com/invite/runpod): Runpod developer community.
- [LinkedIn](https://www.linkedin.com/company/runpod-io/): Runpod company profile.
## Optional
- [AI infrastructure guides](https://www.runpod.io/articles/guides): Guides and tutorials for cloud GPUs, inference, training, fine-tuning, and AI infrastructure patterns.
- [GPU comparison guides](https://www.runpod.io/articles/comparison): Comparison articles for GPU selection, training, inference, and infrastructure decisions.
- [Alternatives articles](https://www.runpod.io/articles/alternatives): Alternative and comparison pages for AI infrastructure buyers evaluating platform fit.
- [GPU Compare](https://www.runpod.io/gpu-compare): GPU benchmark and specification comparison page.
- [Best GPU for AI training](https://www.runpod.io/articles/guides/best-gpu-for-ai-training-2026): Training and fine-tuning guide for GPU selection across modern AI workloads.
- [GPU training vs inference](https://www.runpod.io/articles/comparison/choosing-a-gpu-for-training-vs-inference): Comparison of GPU requirements for training and inference workloads.
- [Run vLLM on Runpod Serverless](https://www.runpod.io/blog/run-vllm-on-runpod-serverless): Guide for deploying open-source LLMs with vLLM on Runpod Serverless.
- [NVIDIA H100](https://www.runpod.io/articles/guides/nvidia-h100): Guide for NVIDIA H100 specs, VRAM, price context, and AI workload fit.
- [NVIDIA H200](https://www.runpod.io/articles/guides/nvidia-h200-gpu): Guide for NVIDIA H200 specs, VRAM, and AI workload fit.
- [NVIDIA B200](https://www.runpod.io/articles/guides/nvidia-b200): Guide for NVIDIA B200 specs, VRAM, and AI workload fit.
- [NVIDIA A100](https://www.runpod.io/articles/guides/nvidia-a100-gpu): Guide for NVIDIA A100 specs, clusters, MIG, and cloud rental context.
- [NVIDIA RTX 4090](https://www.runpod.io/gpu-models/rtx-4090): Runpod GPU model page for NVIDIA RTX 4090 availability and workload fit.
- [NVIDIA RTX 4090 for AI](https://www.runpod.io/articles/guides/nvidia-rtx-4090): Workload-first guide for choosing NVIDIA GPUs for LLMs, fine-tuning, inference, and diffusion workloads.