> For the complete documentation index, see [llms.txt](https://docs.nebulablock.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.nebulablock.com/resources/glossary.md).

# Glossary

Definitions of the terms used across the Nebula Block documentation.

**Inference API** The OpenAI-compatible API for running models — text, vision, image, video, embeddings, and reranking — at `https://inference.nebulablock.com/v1`.

**Platform API** The REST API for managing resources: instances, SSH keys, API keys, object storage, and billing, at `https://api.nebulablock.com/api/v1`.

**Serverless inference** Running a model on managed endpoints, with no servers to provision. Billed on usage.

**GPU instance** A virtual machine, container, or bare-metal server with GPUs attached, rented by the hour.

**Object storage** S3-compatible storage for unstructured data such as datasets, checkpoints, and model outputs.

**Workspace** The container for your object storage. A workspace holds buckets and carries its own S3 access key pair.

**Bucket** A named container for objects inside a storage workspace.

**SSH key** A cryptographic key pair used to log into instances securely. You upload the public half; the private half stays with you.

**API key** A long-lived credential, prefixed `sk-`, that authenticates requests to both APIs.

**Access token** A short-lived JWT issued at login. Suited to interactive sessions rather than applications.

**Tier** Your account level, from Tier 1 to Tier 4. It determines your rate limits, which products you can use, and how much credit you can hold. See [Tiers and Rate Limits](/account/tiers-and-limits.md).

**RPM / TPM / RPD** Requests per minute, tokens per minute, and requests per day — the three rate limits applied to inference calls. RPD counters reset at 00:00 UTC.

**Context length** The maximum number of tokens a model can consider in one request, prompt and response combined.

**Embedding** A dense numeric vector representing text, used for semantic search and retrieval.

**Reranking** Scoring retrieved documents against a query so the most relevant ones can be kept.

**Team** A shared workspace with its own members, roles, and API keys. See [Teams](/account/teams.md).

**Credit** The prepaid balance all usage is deducted from.

**Referral** A program that pays commission on the spend of users you refer.