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

# Providers & Models

> Choose from 105 AI models across 14 providers—or bring your own

## Supported Providers

BoostGPT supports **14 AI providers** with **99 models** spanning from ultra-fast nano models to advanced reasoning models. Thirteen are hosted for you; Ollama runs on your own machine.

> [card] [OpenAI](/docs/providers/openai.md)
>
> 14 models — GPT-6 Astra, GPT-6 Sol, GPT-6 Luna and more

> [card] [Google](/docs/providers/google.md)
>
> 8 models — Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.5 Flash Lite and more

> [card] [Anthropic](/docs/providers/anthropic.md)
>
> 12 models — Claude Fable 5.1, Claude Opus 5.5, Claude Fable 5 and more

> [card] [xAI](/docs/providers/xai.md)
>
> 7 models — Grok 4.7, Grok 4.6, Grok 4.5 and more

> [card] [Zhipu AI](/docs/providers/zhipu.md)
>
> 8 models — GLM-4.6, GLM-4.7, GLM-4.7 Flash and more

> [card] [MiniMax](/docs/providers/minimax.md)
>
> 5 models — MiniMax M3, MiniMax M2.7, MiniMax M2.5

> [card] [DeepSeek](/docs/providers/deepseek.md)
>
> 3 models — DeepSeek V4 Flash, DeepSeek V4 Pro

> [card] [Moonshot AI](/docs/providers/moonshot.md)
>
> 4 models — Kimi K3, Kimi K2.7 Code, Kimi K2.7 Code Highspeed and more

> [card] [Mistral AI](/docs/providers/mistral.md)
>
> 3 models — Mistral Small, Mistral Medium, Mistral Large

> [card] [Cohere](/docs/providers/cohere.md)
>
> 5 models — Command A, Command R+, Command R and more

> [card] [Groq](/docs/providers/groq.md)
>
> 4 models — DeepSeek Llama (70B), GPT-OSS (20B), GPT-OSS (120B) and more

> [card] [Fireworks AI](/docs/providers/fireworks.md)
>
> 11 models — DeepSeek V3, DeepSeek R1, Llama V3 (8B) and more

> [card] [Together AI](/docs/providers/togetherai.md)
>
> 7 models — Qwen 2.5 Coder (32B), Qwen 2.5 Turbo (72B)

> [card] [Ollama](/docs/providers/ollama.md)
>
> 4 models — Llama 3.3 (70B) Local, Llama 3.1 (8B) Local, Llama 3.1 (70B) Local and more

## Key Features

> [accordion] Bring Your Own API Keys
>
> Use your own API keys for OpenAI, Google, Anthropic, xAI, DeepSeek, Mistral, Cohere, Groq, or Ollama. You control costs and rate limits.

> [accordion] Use Our Infrastructure (Optional)
>
> Don't have API keys? Use our pooled infrastructure on paid plans. We handle provisioning, rate limits, and scaling.

> [accordion] Reasoning Models
>
> Access advanced reasoning models like O1, O3 Mini, DeepSeek R1, Gemini Flash Thinking, and Claude Extended Thinking for complex problem-solving.

> [accordion] Smart Model Selection
>
> Choose models based on speed, cost, reasoning capability, and context window. Use fast models for simple queries, reasoning models for complex tasks.

> [accordion] Self-Hosted with Ollama
>
> Run models locally using Ollama. Keep data private, eliminate API costs, and use custom fine-tuned models.

## Model Categories

### Speed-Optimized (Nano/Mini)

Ultra-fast models for simple tasks, high-volume applications, and real-time responses.

**Best for:** FAQs, basic chat, high-traffic bots\
**Examples:** GPT-4.1 Nano, Llama 3.1 Instant (8B), Gemini 2.5 Flash Lite

### Balanced (Standard)

Great all-around models balancing speed, cost, and capability.

**Best for:** Most production use cases, customer support, content generation\
**Examples:** GPT-6 Luna, Claude Sonnet 5.5, Gemini 3.8 Flash

### Advanced (Pro/Large)

Powerful models for complex tasks requiring deep understanding.

**Best for:** Complex queries, creative writing, technical analysis\
**Examples:** GPT-5, Claude Opus 4.1, Gemini 2.5 Pro

### Reasoning Models

Specialized models with extended thinking for complex problem-solving.

**Best for:** Math, coding, logic puzzles, multi-step reasoning\
**Examples:** GPT-6.1 Sol, Claude Opus 5.5, Gemini 3.8 Flash, DeepSeek V4 Pro

[Learn more about reasoning models →](/docs/providers/reasoning-models.md)

## Model Comparison

| Provider | Model | Speed | Cost | Reasoning | Context | Credits per step |
| - | - | - | - | - | - | - |
| OpenAI | GPT-6 Astra | Medium | High | Exceptional | 1.05M | 39 |
| OpenAI | GPT-6.1 Sol | Medium | Medium | Exceptional | 1.05M | 8 |
| OpenAI | GPT-6 Luna | Fast | Low | Excellent | 1.05M | 1 |
| Google | Gemini 3.8 Flash | Fast | Medium | Exceptional | 1.05M | 3 |
| Google | Gemini 3.5 Flash Lite | Very fast | Very low | Good | 1M | 2 |
| Anthropic | Claude Fable 5.1 | Slow | High | Exceptional | 1M | 35 |
| Anthropic | Claude Opus 5.5 | Medium | High | Exceptional | 1M | 15 |
| Anthropic | Claude Sonnet 5.5 | Fast | Medium | Exceptional | 1M | 8 |
| Anthropic | Claude Haiku 4.5 | Fast | Low | Good | 200K | 4 |
| xAI | Grok 4.7 | Medium | Medium | Exceptional | 500K | 9 |
| DeepSeek | DeepSeek V4 Pro | Medium | Low | Excellent | 1M | 5 |
| Groq | GPT-OSS (120B) | Fast | Low | Excellent | 131K | 2 |
| Ollama | Llama 3.3 (70B) Local | Medium | Very low | Excellent | 128K | 0 |

\*Ollama runs on your own host, so it uses no credits; your compute is your own.

[View full comparison →](/docs/providers/model-comparison.md)

## Using Your Own Models (Ollama)

The **only way** to use custom models in BoostGPT is through **Ollama**.

**1. Install Ollama**

Download and install [Ollama](https://ollama.ai) on your server or local machine.

**2. Pull Your Model**

```bash theme={null}
    ollama pull llama3.3:70b
    # Or your custom fine-tuned model
```

**3. Configure BoostGPT**

Set your Ollama host URL in the dashboard or via SDK:

```javascript theme={null}
    const bot = await client.chat({
      bot_id: 'my-bot-id',
      provider_host: 'http://localhost:11434',  // or your server IP
      model: 'llama3.3:70b',
      message: 'Hello, world!' 
    });
```

**4. Deploy**

Your AI teammate will now use your self-hosted Ollama model. No API costs, full control.

[Learn more about Ollama →](/docs/providers/ollama.md)

## Credits System

Your plan includes a monthly credit allowance. Every model call your AI teammate makes uses credits in line with how much they read and write and the model's price, so a short reply on a fast model uses a fraction of a credit and a long AI teammate task on a frontier model uses more.

* **The model picker** shows each model's credits for a typical AI teammate step, so models can be compared at a glance.
* **Each message** shows the tokens it used and the credits it was charged.
* **Your own API keys** (and your own Ollama host) use no credits.

> [!NOTE]
>
> AI teammate turns, plain chat replies, chat titles and other AI calls made on your behalf all use credits. Training data and incoming messages don't.

## Choosing the Right Model

### In the dashboard

1. Go to your bot settings
2. Click "Model Selection"
3. Choose based on your needs:
   * **Fast responses** → GPT-4o Mini, Gemini Flash Lite
   * **Best quality** → GPT-5, Claude Opus, Gemini Pro
   * **Complex reasoning** → O1, DeepSeek R1
   * **Low cost** → Use nano/mini variants

### For Developers (SDK/API)

```javascript theme={null}
const bot = await client.createBot({
  name: 'Support Bot',
  model: 'gpt-6-luna', // Fast & affordable
  // Or for complex queries:
  // model: 'gpt-5.4', // Advanced reasoning
});
```

## Provider-Specific Setup

> [card] [Bring Your Own Keys](/docs/providers/bring-your-own-keys.md)
>
> Use your existing API keys from any provider

> [card] [Use Our Infrastructure](/docs/api-reference/authentication.md)
>
> Let us handle provisioning (paid plans only)

## Next Steps

> [card] [OpenAI](/docs/providers/openai.md)
>
> 14 models — GPT-6 Astra, GPT-6 Sol, GPT-6 Luna and more

> [card] [Google](/docs/providers/google.md)
>
> 8 models — Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash and more

> [card] [Anthropic](/docs/providers/anthropic.md)
>
> 12 models — Claude Fable 5.1, Claude Opus 5.5, Claude Fable 5 and more

> [card] [xAI](/docs/providers/xai.md)
>
> 7 models — Grok 4.7, Grok 4.6, Grok 4.5 and more

> [card] [Zhipu AI](/docs/providers/zhipu.md)
>
> 8 models — GLM-5.3, GLM-5.3 Flash, GLM-5.2 and more

> [card] [MiniMax](/docs/providers/minimax.md)
>
> 5 models — MiniMax M2.7 Highspeed, MiniMax M2.5 Highspeed, MiniMax M3 and more

> [card] [DeepSeek](/docs/providers/deepseek.md)
>
> 3 models — DeepSeek V4 Flash, DeepSeek V4 Flash Vision, DeepSeek V4 Pro

> [card] [Moonshot AI](/docs/providers/moonshot.md)
>
> 4 models — Kimi K3, Kimi K2.7 Code, Kimi K2.7 Code Highspeed and more

> [card] [Mistral AI](/docs/providers/mistral.md)
>
> 3 models — Mistral Small, Mistral Medium, Mistral Large

> [card] [Cohere](/docs/providers/cohere.md)
>
> 5 models — Command A, Command R+, Command R and more

> [card] [Groq](/docs/providers/groq.md)
>
> 4 models — GPT-OSS (20B), Qwen 3.8 (27B), Groq Compound and more

> [card] [Fireworks AI](/docs/providers/fireworks.md)
>
> 11 models — DeepSeek V3, DeepSeek R1, Llama V3 (8B) and more

> [card] [Together AI](/docs/providers/togetherai.md)
>
> 7 models — GLM-5.3, GLM-5.2, Kimi K3 and more

> [card] [Ollama](/docs/providers/ollama.md)
>
> 4 models — Llama 3.3 (70B) Local, Llama 3.1 (8B) Local, Llama 3.1 (70B) Local and more


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.