# Free models

> The free model, its rate limit, and what you give up.

Source: https://reverie.im/en/docs/reference/free-models
Language: en

There is exactly one free model: **Llama 3.1 8B**. Selecting it in **Choose AI Model** costs no credits at all. What it costs instead is a message quota and several features.

## The model \[#the-model]

| Property   | Value                                                                                            |
| ---------- | ------------------------------------------------------------------------------------------------ |
| Multiplier | 0× — no credits are charged, ever                                                                |
| Context    | 131,072 tokens, the smallest of any selectable model                                             |
| Reasoning  | None                                                                                             |
| Vision     | None — it can't read images you attach                                                           |
| Tools      | None — see [What you give up](https://reverie.im/en/docs/reference/free-models#what-you-give-up) |
| Tier       | Basic, so no subscription is needed to select it                                                 |

## The quota \[#the-quota]

| Fact     | Value                                                                                            |
| -------- | ------------------------------------------------------------------------------------------------ |
| Messages | **15 per 3 hours**                                                                               |
| Window   | Rolling, not a daily reset — each message frees its own slot 3 hours after you sent it           |
| Scope    | Per account, **shared** across web chat, novels, the Discord and Telegram bots, and the Open API |
| Pro      | ×3 → 45 messages per 3 hours                                                                     |
| Premium  | ×4 → 60 messages per 3 hours                                                                     |
| Ultimate | ×5 → 75 messages per 3 hours                                                                     |

One pool means bot and API traffic eat into the same allowance as your web chats. When it runs out, the send is refused and you're told when to retry; through the [Open API](https://reverie.im/en/docs/integrations/open-api) the same condition comes back as HTTP 429 with a `Retry-After` header.

A subscription raises this quota, but a plan's monthly credits already cover turns on models that read images, call plugins and hold far more context. Treat the higher free-model quota as a fallback for the end of a billing period, not as the reason to subscribe.

## What you give up \[#what-you-give-up]

The free model has no tool-calling support. In a chat on it:

* **Plugins that need tool calls don't run.** The plugin card marks which ones require them. [Using plugins](https://reverie.im/en/docs/plugins/using-plugins)
* **World-book lore lookup doesn't run.** Only always-on entries reach the model; nothing is retrieved on demand. [World books](https://reverie.im/en/docs/writing/world-books)
* **The character can't save or update memories mid-chat.** Memories you write yourself in the **Memory** tab still work. [Memory](https://reverie.im/en/docs/chatting/memory)

Beyond tools:

* **No image input.** Attaching a picture to a message needs a vision model — the cheapest is GLM 5.3 Flash at 0.5×.
* **No reasoning mode.**
* **The smallest context window.** At 131,072 tokens a long chat starts leaning on summaries sooner than it would on any other model.
* **[Story mode](https://reverie.im/en/docs/writing/story-mode) doesn't offer it** — stories only list models that support tool calls. Novels do offer it.
* **It never appears as the alternative** in the two-reply comparison card, which only draws from models that cost credits.

Everything else works normally: forking, personas, scenarios, group chats, voice calls, stickers and image generation are unaffected by which text model you're on. Image generation and voice are billed at their own rates whatever the chat model.

> **Chatting without an account is a separate allowance.** Anonymous chats get 5 messages, or 8 after the verification check, renewed 24 hours after the first one and capped at 30 per day per IP address, and they run on the default paid model rather than this one. [Limits](https://reverie.im/en/docs/reference/limits)

## Other ways to pay nothing \[#other-ways-to-pay-nothing]

Credits also arrive without payment: a 30,000-credit grant when you sign up, plus [daily check-in](https://reverie.im/en/docs/rewards/daily-check-in) credits and the [lucky draw](https://reverie.im/en/docs/rewards/lucky-draw). Spent on the 0.3× default at roughly 2,400 credits a turn, they buy turns with tools, memory writes and a larger context window — which the free model can't give you at any quota. [Credits](https://reverie.im/en/docs/billing/credits)

## Related \[#related]

**[Models](https://reverie.im/en/docs/reference/models)**

All eleven selectable models with their multipliers and capabilities.

**[Choosing a model](https://reverie.im/en/docs/reference/choosing-a-model)**

What to switch to when the free model isn't enough.

**[Credits](https://reverie.im/en/docs/billing/credits)**

How a turn is priced and where free credits come from.

**[Limits](https://reverie.im/en/docs/reference/limits)**

Every quota and cap in one table.