Cerebras Inference

Low-latency inference for open-weight text and vision models on wafer-scale accelerators.

Provider: default
Kind: AI service
Category: Compute
Official docs: https://inference-docs.cerebras.ai/support/rate-limits
Also known as: Cerebras Cloud

Properties · 6 set by this component

property	type	values	required	kind default	this component
auth	enum	api-key \| oauth \| iam \| none	no	—	▸ api-key
cachedTokensCountToUncachedQuota · component property	boolean	—	no	false	inherited
contentFiltering	boolean	—	no	—	inherited
customerManagedKeys	boolean	—	no	—	inherited
dataResidency · advanced	enum	us \| eu \| asia \| global	no	—	inherited
dataRetentionDays	integer	—	no	—	inherited
keyStore	boolean	—	no	false	inherited
managed	boolean	—	no	false	▸ true
modality	enum	text \| vision \| speech \| multimodal \| structured \| image \| video \| audio \| embedding \| rerank	no	—	▸ multimodal
publicEndpoint	boolean	—	no	—	▸ true
quotaUnitsPerMinute	integer	—	no	—	inherited
rateLimitRpm · advanced	integer	—	no	—	▸ 5
rateLimitTpm · advanced	integer	—	no	—	▸ 30000
serving · component property	enum	shared \| dedicated	no	shared	inherited
tier · component property	enum	free-trial \| developer \| enterprise	no	free-trial	inherited
viaGateway	boolean	—	no	false	inherited

Rules that apply · 2

Every rule whose selector covers this component's kind or its category. Whether a rule fires depends on the design: its properties, its edges and its groups.

AI API key not managedwarning

ai-api-key-unmanaged

This AI vendor is reached with an API key, and nothing in the design holds or rotates that key. It ends up in environment files or code, every caller shares it, and a single leak runs up the vendor bill until someone notices. Put the key in a key-management component, or say it is held there with keyStore.

— all of these hold:
— the node's `auth` is "api-key"
— the node's `keyStore` is false
— the node has at least one incoming edge from an app-server, worker or client
— this does not hold:
— the design contains a node of kind key-management

Stateful node without zone redundancyerror

stateful-node-without-zone-redundancy

A stateful compute node pinned to exactly one availability zone disappears with that zone, and its state goes with it. Nodes inside a cell are exempt, because there the cell is deliberately single-zone and redundancy comes from the other cells.

— all of these hold:
— the node's `stateless` is false
— the node sits inside exactly 1 availability-zone group
— this does not hold:
— the node sits inside at least one cell group

Use it in a design

nodes:
  - { id: "cerebras-api", component: "cerebras-api" }

Other AI service products · 125

and 85 more → AI service page

Source

1# yaml-language-server: $schema=../../schema/component.schema.json
2# Cerebras Inference: hosted open-weight text and vision models on wafer-scale chips over a public endpoint.
3# Per "Rate Limits", limits are per organization and model (RPM, TPM, TPH, TPD); the Free Trial allows 5 RPM
4# and 30K uncached TPM, and total TPM is 3x uncached so cached tokens only count toward the total bucket.
5schemaVersion: "1"
6type: "component"
7id: "cerebras-api"
8kind: "ai-service"
9provider: "default"
10docs: "https://inference-docs.cerebras.ai/support/rate-limits"
11aliases: ["Cerebras Cloud"]
12defaults:
13  auth: "api-key"
14  managed: true
15  modality: "multimodal"
16  publicEndpoint: true
17  rateLimitRpm: 5
18  rateLimitTpm: 30000
19properties:
20  tier:
21    type: "enum"
22    values: ["free-trial", "developer", "enterprise"]
23    default: "free-trial"
24  serving:
25    type: "enum"
26    values: ["shared", "dedicated"]
27    default: "shared"
28  cachedTokensCountToUncachedQuota:
29    type: "boolean"
30    default: false

The same page as markdown, for an agent: /catalogue/components/cerebras-api.md