## Overview

The Models page is your comprehensive AI model catalog. Browse, search, and compare 50+ models from leading providers like Anthropic, OpenAI, Google, DeepSeek, and more. View real-time pricing, capabilities, and performance specs to choose the right model for your use case.

## Key Features

<CardGroup cols={2}>
  <Card title="Model Discovery" icon="search">
    Search and filter through all available models by provider, tier, or name.
  </Card>
  <Card title="Real-Time Pricing" icon="payments">
    View current pricing with discounts applied. See input, output, and cache
    pricing.
  </Card>
  <Card title="Capability Comparison" icon="fact_check">
    Compare model capabilities: tool use, vision, extended thinking, streaming,
    and more.
  </Card>
  <Card title="Performance Specs" icon="speed">
    View context window size, max output, latency ratings, and data cutoff
    dates.
  </Card>
</CardGroup>

## Using the Models Page

### Searching for Models

Use the search bar at the top to find models by name or provider:

**Example searches:**

- `claude` - Shows all Claude models
- `gpt-4o` - Shows GPT-4o variants
- `anthropic` - Shows all Anthropic models
- `flash` - Shows fast models (Gemini Flash, etc.)

The search is case-insensitive and updates results in real-time as you type.

### Filtering by Provider Family

Click the provider buttons to filter by model family:

- **All Families** - Show all models
- **Anthropic** - Claude models only
- **OpenAI** - GPT models only
- **Google** - Gemini models only
- **DeepSeek** - DeepSeek models only
- And more...

Each button shows the provider icon for quick visual identification.

### Filtering by Tier

Filter models by their pricing tier:

| Tier         | Description                              | Examples                                 |
| ------------ | ------------------------------------------ | ------------------------------------------- |
| **Economy**  | Deepest prepaid discount, no bonus discount | GPT-4o-mini, Claude Haiku, Gemini Flash    |
| **Standard** | Discounted models for everyday use        | GPT-4o-mini, Claude Haiku, Gemini Flash    |
| **Premium**  | Full-price, most capable frontier models  | Claude Opus, GPT-5.5                       |

**Tier Benefits:**

Each tier has a different rate for your **bonus** wallet and your **prepaid** wallet — they are not the same number. See [Provider Tiers](/docs/models/provider-tiers) for the full breakdown, including how a single request can split across both wallets.

- Economy: prepaid 0.2x by default (80% off provider cost); bonus 1.0x (no discount); payable from bonus or prepaid wallet balance
- Standard: prepaid 0.5x by default (50% off provider cost, less for some model families); bonus 1.5x (50% markup); payable from bonus or prepaid wallet balance
- Premium: 1.0x (full provider price, no discount, no bonus/prepaid distinction since bonus can't pay it); payable from prepaid wallet balance only

<Info>
  Only models available on your enabled tiers are fully highlighted. Disabled
  tier models appear dimmed.
</Info>

### Sorting Options

Sort the model list by different criteria:

| Sort By     | Purpose                      | Use When                                   |
| ----------- | ---------------------------- | ------------------------------------------ |
| **Name**    | Alphabetical order           | Finding a specific model quickly           |
| **Latency** | Fastest response first       | Speed is critical for your use case        |
| **Context** | Largest context window first | Processing long documents or conversations |
| **Price**   | Cheapest input tokens first  | Budget optimization is your priority       |

Click the sort dropdown and select your preferred option.

## Understanding Model Cards

Each model card displays comprehensive information:

### Header Section

- **Provider Icon** - Visual indicator of model family
- **Model Name** - Display name (e.g., "Claude Sonnet 4.6")
- **Model ID** - API identifier (e.g., `claude-sonnet-4-6`)
  - Click the ID or copy button to copy to clipboard
  - Use this ID in your API requests
- **Discount Badge** - Shows discount percentage if model has a discount applied

### Description

- Brief description of the model
- "Strongest for" use case (e.g., "Long context reasoning", "Fast generation")

### Specifications

**Provider Tier Badges:**
Shows which tiers this model is available on:

- **Filled badge** = Tier is enabled in your settings
- **Outlined badge** = Tier is available but not enabled

**Context & Output:**

- **Context Window** - Maximum input size (e.g., "128K" = 128,000 tokens)
- **Max Output** - Maximum response size (e.g., "16K" = 16,000 tokens)

**Capabilities:**
Visual badges showing model capabilities:

- **Tool Use** - Supports function/tool calling
- **Extended Thinking** - Internal reasoning display (Claude Opus, o1)
- **Prompt Caching** - Supports prompt caching for cost savings
- **Streaming** - Supports real-time streaming responses
- **Vision** - Can process images (GPT-4o, Claude Sonnet, Gemini)

### Pricing Grid

Four types of pricing displayed:

| Price Type      | What It Means                  | Example            |
| --------------- | ------------------------------ | ------------------ |
| **Input**       | Cost per million input tokens  | $3.00 / 1M tokens  |
| **Output**      | Cost per million output tokens | $15.00 / 1M tokens |
| **Cache Read**  | Cost to read from prompt cache | $0.30 / 1M tokens  |
| **Cache Write** | Cost to write to prompt cache  | $3.75 / 1M tokens  |

**Discount Indicators:**

- Original price shown with strikethrough
- Discounted price in green
- Discount percentage badge in header

## Use Cases

### Finding the Right Model

**Scenario 1: Cost Optimization**

1. Sort by **Price** (cheapest first)
2. Filter by **Standard** tier
3. Compare capabilities to ensure model meets your needs
4. Copy model ID and use in your code

**Scenario 2: Long Document Processing**

1. Sort by **Context** (largest first)
2. Look for models with 128K+ context window
3. Check if model has "Tool Use" if you need structured extraction
4. Use models like `claude-sonnet-4-6` (200K context)

**Scenario 3: Fast Responses**

1. Sort by **Latency** (fastest first)
2. Filter by **Standard** tier
3. Choose models marked for fast generation
4. Examples: `gemini-2.5-flash`, `gpt-4o-mini`

**Scenario 4: Vision Tasks**

1. Search for models with "Vision" capability badge
2. Compare pricing (vision models are typically more expensive)
3. Examples: `gpt-4o`, `claude-sonnet-4-6`, `gemini-2.5-pro`

### Comparing Models

**Side-by-Side Comparison Example:**

| Feature                       | GPT-4o-mini                      | Claude Sonnet 4.6                     | Gemini 2.5 Flash          |
| ------------------------------ | --------------------------------- | --------------------------------------- | -------------------------- |
| **Standard, Prepaid Rate**    | 0.2x (OpenAI family override)     | 0.35x (Anthropic family override)      | 0.5x (default)             |
| **Standard, Bonus Rate**      | 1.5x (no family overrides)        | 1.5x (no family overrides)             | 1.5x (default)             |
| **Economy, Prepaid Rate**     | 0.2x (default, no overrides yet)  | 0.2x (default, no overrides yet)       | 0.2x (default)             |
| **Context**              | 128K                             | 200K                                 | 1M                       |
| **Max Output**   | 16K                      | 8K                | 8K                       |
| **Input Price**  | $0.15/1M                 | $3.00/1M          | $0.15/1M                 |
| **Output Price** | $0.60/1M                 | $15.00/1M         | $0.60/1M                 |
| **Tool Use**     | ✅                       | ✅                | ✅                       |
| **Vision**       | ✅                       | ✅                | ✅                       |
| **Streaming**    | ✅                       | ✅                | ✅                       |
| **Best For**     | High-volume simple tasks | Complex reasoning | Long context at low cost |

## Tips & Best Practices

### Maximizing Your Budget

1. **Use Economy or Standard tier** when possible - gets a prepaid discount instead of Premium's full price (Economy also works from bonus at full face value; Standard's bonus rate is actually a markup)
2. **Enable prompt caching** for repeated prompts - Cache reads cost 10x less
3. **Match model to task complexity** - Don't use Opus for simple tasks
4. **Monitor tier rates** - on prepaid, Economy is 0.2x by default, Standard is 0.5x by default (less for some model families), Premium is always 1.0x; on bonus, Economy is 1.0x, Standard is 1.5x, and Premium isn't payable from bonus at all

### Model Selection Guidelines

| Task Type                       | Recommended Tier | Example Models                          |
| -------------------------------- | ----------------- | ---------------------------------------- |
| Bulk/background, cost-sensitive | Economy           | GPT-4o-mini, Claude Haiku, Gemini Flash |
| Simple Q&A, summarization       | Standard          | GPT-4o-mini, Claude Haiku, Gemini Flash |
| Code generation, analysis       | Standard         | Claude Sonnet, Gemini Pro                |
| Complex reasoning, architecture | Premium          | Claude Opus, GPT-5.5                     |
| Long document processing        | Standard/Premium | Claude Sonnet (1M), Gemini Pro (2M)     |
| Vision tasks                    | Standard/Premium | GPT-4o-mini, Claude Sonnet, Gemini Pro  |

### Understanding Discounts

Models with discount badges show **platform-wide discounts** applied by ModelStack:

**Example:**

- Original: $3.00/1M input tokens
- Discount: 10% off
- Your price: $2.70/1M input tokens

**This is in addition to your tier's wallet rate — and prepaid and bonus differ:**

- Economy, prepaid: $2.70 × 0.2 = $0.54 (default Economy prepaid rate)
- Standard, prepaid: $2.70 × 0.5 = $1.35 (default Standard prepaid rate)
- Standard, bonus: $2.70 × 1.5 = $4.05 (Standard bonus rate is a markup, not a discount)
- Premium: $2.70 × 1.0 = $2.70 (full price, no discount, prepaid only)

<Info>
  A single request paid partly from bonus and partly from prepaid lands
  somewhere between these numbers — see [How a Request Can Split Across
  Wallets](/docs/models/provider-tiers#how-a-request-can-split-across-wallets)
  for a full worked example.
</Info>

### Enabling More Tiers

By default, **Economy** and **Standard** tiers are enabled. To access Premium models:

1. Go to **Dashboard → Settings**
2. Navigate to **Model Routing** tab
3. Enable Premium tier
4. Return to Models page - newly enabled models will be highlighted

## Common Questions

<AccordionGroup>
  <Accordion title="Why are some models dimmed?">
    Models appear dimmed when they're available on tiers you haven't enabled in
    your settings. Go to **Settings → Model Routing** to enable more tiers.
  </Accordion>
  <Accordion title="What does 'Cache Read/Write' pricing mean?">
    Prompt caching lets you cache repeated prompts (like system instructions) to
    save costs. Cache writes store the prompt, cache reads retrieve it. Cache
    reads cost ~10x less than regular input tokens.
  </Accordion>
  <Accordion title="Can I request new models?">
    Yes! Contact support or submit a feature request through the dashboard. We
    regularly add new models based on user demand.
  </Accordion>
  <Accordion title="How often is pricing updated?">
    Pricing is updated in real-time when providers announce changes. Discounts
    are applied automatically - no action needed.
  </Accordion>
  <Accordion title="What's the difference between context and max output?">
    **Context window** is the total input size (your prompt + conversation
    history). **Max output** is how much the model can generate in a single
    response. For example, Claude Sonnet has 200K context but 8K max output.
  </Accordion>
</AccordionGroup>

## Related Documentation

- [Provider Tiers](/docs/models/provider-tiers) - Understand Economy, Standard, and Premium tiers
- [Supported Models](/docs/models/supported-models) - Complete model catalog with specifications
- [Billing & Balance](/docs/billing/balance) - How wallet balance and tier multipliers work
- [API Reference](/docs/api-reference/models) - Using models in your API requests