Matrix OSMatrix OS

Matrix AI models

Explore the managed Workers AI model catalog, the GLM Flash default, rollout availability, and how ordinary Chat and Matrix Bot share credit.

Matrix AI is managed access to models from your Matrix computer. You use Matrix AI credit instead of adding a personal model-provider key.

Expanded catalog rollout

Eight additional Cloudflare Workers AI models are being prepared for rollout. The list below describes the planned managed catalog; it does not mean every model is available on your computer today. Your current model picker and Settings → Agents & providers show the models your computer can use.

Default and model choice

GLM 5.3 Flash remains the default for new managed Matrix AI Chats when it is available. Adding models preserves existing saved choices.

Flash has a low published price compared with several larger options below. A default is a starting point: compare models on your own task before choosing one for ongoing work. OpenRouter's rankings measure usage, which is different from a task-specific quality evaluation.

Managed Workers AI catalog

All nine entries use Cloudflare Workers AI through Matrix AI. This expansion covers text and tool use. Support for a model in Cloudflare's catalog does not by itself make it available through Matrix.

The prices below are Cloudflare's published reference rates in USD per million tokens, checked on October 15, 2026. They are not a separate Matrix price quote. Cached input can have a different rate; consult Cloudflare's pricing page for the current details. Your Matrix credit balance and recorded usage remain the source for your account's charges.

ModelInputOutput
GLM 5.3 Flash — default$0.15$0.50
DeepSeek V4 Flash 0731$0.44$1.32
GLM 5.3$1.40$4.40
Kimi K2.7 Code$0.95$4.00
GPT-OSS 120B$0.35$0.75
Gemma 4 26B A4B$0.10$0.30
DeepSeek V4 Pro 0813$1.32$3.96
Qwen 3.8 27B$0.45$3.20
GPT-OSS 20B$0.20$0.30

Exact model identifiers

Model versions matter. For example, DeepSeek V4 Flash 0731 is a different selection from DeepSeek V4.1 Flash. These identifiers match the planned managed catalog and Cloudflare's model catalog.

@cf/zai-org/glm-5.3-flash
@cf/deepseek-ai/deepseek-v4-flash-0731
@cf/zai-org/glm-5.3
@cf/moonshotai/kimi-k2.7-code
@cf/openai/gpt-oss-120b
@cf/google/gemma-4-26b-a4b-it
@cf/deepseek-ai/deepseek-v4-pro-0813
@cf/qwen/qwen3.8-27b
@cf/openai/gpt-oss-20b

Availability on your computer

A model becomes usable only after the supporting Matrix update and model rollout reach your computer, your account is eligible for that exact model, and Matrix can freshly verify the model's service readiness. One working model does not establish that another model is ready.

Ordinary Chat and Matrix Bot use the same managed model access and funding checks. Your selected computer's current catalog determines which models can run. Keep your app and Matrix computer updated; after an update, reopen Settings and choose Refresh provider status to check the current state.

If Matrix AI access needs reconnecting, support may provide a reviewed Matrix OS repair update for the affected computer. Finish active work before any requested restart, then refresh provider status. Installing an update alone is not proof of restored access: Matrix must confirm fresh availability and readiness for each model. A repair does not add credit or reset usage.

Credit and usage

Changing models uses the same Matrix AI Chat credit and monthly spending limits. It does not grant new credit or reset usage. Ordinary Chat and Matrix Bot both consume the applicable owner credit when they use managed models.

Matrix reserves a bounded amount while a funded request runs, then confirms the charge from its final reported usage. Credit can remain reserved while usage is still being confirmed. A displayed balance alone does not guarantee that a particular model is ready or that the full balance is currently spendable.

See Matrix AI Chat credit for available credit, usage history, reserved credit, and recovery guidance.

How is this guide?

On this page