> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fife.bot/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Models

> Understanding model tiers, credit costs, and how to choose the right model for your chatbot.

FIFE.BOT gives you access to models from OpenAI, Anthropic, Google, Meta, and others. Models are grouped into **5 tiers** by capability and credit cost.

## Model Tiers

<Tabs>
  <Tab title="Free">
    **0 credits per message** — included on every plan.

    | Model        | Provider | Strengths                                                    |
    | ------------ | -------- | ------------------------------------------------------------ |
    | GPT-OSS 120B | OpenAI   | Capable open-weight stack for straightforward Q\&A           |
    | GPT-4o Mini  | OpenAI   | Fast answers, strong instruction following at no credit cost |

    <Note>
      Free-tier models handle FAQs and light tasks well. For harder reasoning or richer writing, use a paid tier on Pro or Business.
    </Note>
  </Tab>

  <Tab title="Budget">
    **1 credit per message** — **Pro** and **Business** only.

    | Model            | Provider | Strengths                                              |
    | ---------------- | -------- | ------------------------------------------------------ |
    | Llama 4 Maverick | Meta     | General chat, solid multilingual coverage              |
    | Gemini 2.0 Flash | Google   | Very fast, great for structured or high-volume replies |
    | DeepSeek V3      | DeepSeek | Strong value for technical and analytical prompts      |
    | Qwen3 235B       | Alibaba  | Reliable general-purpose assistant behavior            |

    <Tip>
      **Gemini 2.0 Flash** or **Llama 4 Maverick** are strong defaults when you outgrow free models but still want low cost per message.
    </Tip>
  </Tab>

  <Tab title="Standard">
    **5 credits per message** — **Pro** and **Business** only.

    | Model            | Provider  | Strengths                                          |
    | ---------------- | --------- | -------------------------------------------------- |
    | Claude Haiku 4.5 | Anthropic | Quick, careful replies with good tone control      |
    | Gemini 2.5 Pro   | Google    | Balanced quality for longer explanations and tools |
  </Tab>

  <Tab title="Premium">
    **10 credits per message** — **Pro** and **Business** only.

    | Model           | Provider  | Strengths                                                       |
    | --------------- | --------- | --------------------------------------------------------------- |
    | Claude Sonnet 4 | Anthropic | Nuanced writing, strong reasoning on complex topics             |
    | GPT-4o          | OpenAI    | Broad knowledge, great for mixed technical and business content |

    <Warning>
      A Pro plan (2,000 credits/month) allows about **200** messages at this tier if every message uses only Premium-priced models.
    </Warning>
  </Tab>

  <Tab title="Ultra">
    **22 credits per message** — **Business** only.

    | Model             | Provider  | Strengths                                                  |
    | ----------------- | --------- | ---------------------------------------------------------- |
    | GPT 5.3           | OpenAI    | Maximum depth for hard problems and long chains of thought |
    | Claude Sonnet 4.6 | Anthropic | Latest Sonnet generation for top-tier quality              |

    <Warning>
      Ultra is meant for high-value conversations. It is not available on the Free or Pro plan.
    </Warning>
  </Tab>
</Tabs>

## Choosing the Right Model

### By Budget

| Priority                  | Recommended Tier   |
| ------------------------- | ------------------ |
| No budget / testing       | Free               |
| Cost-conscious production | Budget             |
| Quality over cost         | Standard or higher |

### By Question Complexity

| Complexity        | Examples                                       | Tier                |
| ----------------- | ---------------------------------------------- | ------------------- |
| Simple FAQ        | "What are your hours?", "How do I return?"     | Free or Budget      |
| Moderate          | "Compare your plans", "Walk me through setup"  | Budget or Standard  |
| Complex reasoning | "Analyze this clause", "Debug this issue"      | Standard or Premium |
| Expert-level      | "Review for compliance", "Multi-step analysis" | Premium or Ultra    |

### By Use Case

<Columns cols={2}>
  <Card title="Customer Support" icon="headset">
    **GPT-4o Mini** (Free) or **Gemini 2.0 Flash** (Budget) — Fast, clear answers for common tickets.
  </Card>

  <Card title="Sales Assistant" icon="handshake">
    **Claude Sonnet 4** (Premium) — Natural, persuasive copy while staying on-brand.
  </Card>

  <Card title="Internal Knowledge Base" icon="building">
    **GPT-4o Mini** (Free) or **Llama 4 Maverick** (Budget) — Good accuracy for policies at scale.
  </Card>

  <Card title="Technical Docs" icon="code">
    **GPT-4o** (Premium) — Code snippets, APIs, and error explanations.
  </Card>

  <Card title="Legal / Compliance" icon="scale-balanced">
    **Claude Sonnet 4** (Premium) or **Ultra** models — Thorough review on **Business** when stakes are highest.
  </Card>

  <Card title="Simple Landing Page FAQ" icon="circle-question">
    **GPT-OSS 120B** or **GPT-4o Mini** (Free) — Short answers from a compact knowledge base.
  </Card>
</Columns>

## Credits Math

How many messages each plan supports per tier (approximate, if you used only that tier):

| Plan     | Credits | Free (0)  | Budget (1) | Standard (5) | Premium (10) | Ultra (22) |
| -------- | ------- | --------- | ---------- | ------------ | ------------ | ---------- |
| Free     | 50      | Unlimited | —          | —            | —            | —          |
| Pro      | 2,000   | Unlimited | 2,000      | 400          | 200          | —          |
| Business | 5,000   | Unlimited | 5,000      | 1,000        | 500          | 227        |

Entries marked **—** mean that tier is not on your plan (see below).

## Changing Models

You can switch your chatbot's model anytime in **Settings → Model & Instructions**. The change takes effect immediately for new messages — existing conversations keep using the model they started with.

<Tip>
  Start with **Free** or **Budget**, watch quality in Analytics, then move up only if you need it. You can buy extra credits on the Billing page.
</Tip>

## Model Access by Plan

Which tiers each subscription can use:

| Tier                                                                            | Free plan | Pro plan | Business plan |
| ------------------------------------------------------------------------------- | --------- | -------- | ------------- |
| **Free models** (GPT-OSS 120B, GPT-4o Mini)                                     | Yes       | Yes      | Yes           |
| **Budget models** (Llama 4 Maverick, Gemini 2.0 Flash, DeepSeek V3, Qwen3 235B) | No        | Yes      | Yes           |
| **Standard models** (Claude Haiku 4.5, Gemini 2.5 Pro)                          | No        | Yes      | Yes           |
| **Premium models** (Claude Sonnet 4, GPT-4o)                                    | No        | Yes      | Yes           |
| **Ultra models** (GPT 5.3, Claude Sonnet 4.6)                                   | No        | No       | Yes           |