> ## Documentation Index
> Fetch the complete documentation index at: https://www.rumus.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Built-in models
> Models included with Rumus Pro, what each one costs, and how to track your usage.
The fastest way to use AI in Rumus is the **built-in models** — a curated set of models from leading labs, available the moment you sign in on a Pro subscription. No upstream API key to manage, no separate provider account, and everything is billed on your Rumus account.
## Plan
There's a single subscription:
| Plan | Price | What you get |
| -------- | ------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Free** | \$0 | All local terminal features. **No** built-in AI models. You can still bring your own provider key — see [Models & Providers](/docs/models/overview). |
| **Pro** | \$10 / month | Cloud sync, encrypted vault sync, built-in web search, **access to every built-in AI model**, and **\$5 of monthly AI credits** included. Usage above the credit is pay-as-you-go. |
Manage your subscription at [rumus.ai](https://www.rumus.ai) → Account, or in the app at **Settings → Account**.
## How billing works
When you call a built-in model, your account is charged for the **actual provider cost** of that request plus a small Rumus token fee on top. The token fee covers routing, billing, support, and the infrastructure that keeps the built-in models working without you managing keys.
## Supported models
The full list, kept in sync with what's exposed at **Settings → AI → Models**. Prices are **per 1M tokens**, in USD.
### Anthropic
| Model | Context | Input | Output | Cache read | Cache write | Vision | Tools | Privacy |
| ----------------- | ------- | ------ | ------- | ---------- | ----------- | ------ | ----- | ------- |
| Claude Opus 4.7 | 1M | \$5.00 | \$25.00 | \$0.50 | \$6.25 | ✓ | ✓ | ✓ |
| Claude Sonnet 4.6 | 1M | \$3.00 | \$15.00 | \$0.30 | \$3.75 | ✓ | ✓ | ✓ |
| Claude Haiku 4.5 | 200K | \$1.00 | \$5.00 | \$0.10 | \$1.25 | ✓ | ✓ | ✓ |
### OpenAI
| Model | Context | Input | Output | Cache read | Vision | Tools | Privacy |
| ------------- | ------- | ------- | -------- | ---------- | ------ | ----- | ------- |
| GPT-5.4 | 1.1M | \$2.50 | \$15.00 | \$0.25 | ✓ | ✓ | – |
| GPT-5.3 Codex | 400K | \$1.75 | \$14.00 | \$0.17 | ✓ | ✓ | – |
| GPT-5.2 Pro | 2M | \$21.00 | \$168.00 | – | ✓ | ✓ | – |
### Google
| Model | Context | Input | Output | Cache read | Vision | Tools | Privacy |
| -------------------- | ------- | ------ | ------- | ---------- | ------ | ----- | ------- |
| Gemini 3 Pro Preview | 1M | \$2.00 | \$12.00 | \$0.20 | ✓ | ✓ | ✓ |
| Gemini 3 Flash | 1M | \$0.50 | \$3.00 | \$0.05 | ✓ | ✓ | ✓ |
### Z.ai
| Model | Context | Input | Output | Cache read | Tools | Privacy |
| ------------ | ------- | ------ | ------ | ---------- | ----- | ------- |
| GLM-5.1 | 203K | \$1.40 | \$4.40 | \$0.26 | ✓ | – |
| MiniMax M2.7 | 205K | \$0.30 | \$1.20 | \$0.06 | ✓ | ✓ |
### DeepSeek
| Model | Context | Input | Output | Cache read | Tools | Privacy |
| ----------------- | ------- | ------ | ------ | ---------- | ----- | ------- |
| DeepSeek V4 Pro | 1M | \$1.74 | \$3.48 | \$0.14 | ✓ | – |
| DeepSeek V4 Flash | 1M | \$0.14 | \$0.28 | \$0.01 | ✓ | – |
### Moonshot AI
| Model | Context | Input | Output | Cache read | Tools | Privacy |
| --------- | ------- | ------ | ------ | ---------- | ----- | ------- |
| Kimi K2.6 | 262K | \$0.95 | \$4.00 | \$0.16 | ✓ | – |
**Privacy** column: a check means the upstream provider has committed to not retaining or training on your data. Models without a check still send only the request itself — Rumus never logs the contents of your prompts or completions.
### How to read the price columns
* **Input / Output** — what you send and what the model generates.
* **Cache read** — input tokens served from the prompt cache, billed cheaper than a fresh input. Long stable system prompts (skills, rules, big files) benefit the most.
* **Cache write** — input tokens being cached for future calls. Anthropic models price cache writes higher than fresh inputs; other providers cache automatically with no separate write price.
* **Reasoning tokens** — for models that "think" before answering, billed at the **Output** rate.
Prices change as upstream providers update their pricing. The in-app table at **Settings → AI → Models** is always authoritative.
## How much usage do I need?
A rough guide to help you choose between staying on the included credit and topping up:
| Profile | Typical monthly usage |
| ------------------------------------------ | --------------------- |
| Daily user, mostly Tab autocomplete | Under \$20 |
| Light Agent user | \$20 |
| Daily Agent user | $60 – $100 |
| Power user (multiple Agents / automations) | \$200+ |
These are observations, not limits. Your actual cost depends on which models you pick and how long the conversations are.
## Monthly credits and overage
Pro includes **\$5 of AI credits each month**. They're spent first, automatically. Once they're used up, requests continue against your account balance at the per-token prices above (plus the Rumus token fee).
**Unused credits don't roll over** — the \$5 resets at the start of each billing cycle.
There's no automatic alerting when you cross a threshold. If you want to keep an eye on usage, check the dashboard regularly (see below).
## Top up your balance
If you expect to use more than the included credits, pre-load your account:
In the app, go to **Settings → Account**.
Next to your **Balance**, click the **Top up** button.
Pick one of the preset top-up amounts and complete the payment. Once it clears, the credit lands in your balance and is immediately usable. You'll get a receipt by email.
## View your balance, usage, and bills
Two places to check what you've spent:
Your current balance, monthly credit remaining, and usage charts (by day, by model).
The detailed per-request log: timestamp, model, token counts, base provider cost, the Rumus token fee, and total.
Token counts are also visible on each AI message in the sidebar — click the small info icon to see exactly what that one request cost.
## Privacy of built-in model traffic
Requests to built-in models are routed through Rumus's API on the way to the upstream provider. We:
* **Do** log token counts, model IDs, and timing for billing.
* **Do not** log message contents or completion text.
* Highlight models from providers with no-retention / no-training commitments using the privacy column above.
For a deeper look at how local secrets are stored, see [Vault & encryption](/docs/account/vault).
## FAQ
A small percentage on top of the base provider cost. The exact rate and the amount per request are shown in the in-app pricing UI and on each line of your bill, so there's no guessing — every charge is itemized.
Yes. Stay on Free, or stay on Pro for sync without using the built-in models — just add your own provider in **Settings → AI → Models**. Built-in models simply won't be picked unless you choose them.
No. The \$5 monthly credit resets at the start of each billing cycle.
Some upstream providers cache transparently with no separate cache-read price (you pay full input rate either way), while a few don't cache at all. The dash means there's no separate cache pricing to show — not that the model is broken.
Billing or model question we didn't cover? Ask in the [Rumus community](https://www.rumus.ai/community).
## Next steps
Use Anthropic, OpenAI, Google, Z.AI, DeepSeek, Kimi, Ollama, or any OpenAI-compatible endpoint.
What you can actually do with the agent now that a model is set up.