Model Selection
Outter provides flexible model selection, allowing you to choose the best AI engine for your use case. Whether you need a general-purpose LLM, a domain-specific model, or an efficient self-hosted solution, Outter ensures seamless integration across multiple providers.
How It Works
Outter is provider-agnostic, meaning you can dynamically switch the underlying AI engine per request. When making an API call, you can specify a model such as OpenAI’s gpt-4o, Anthropic’s claude-3.7, or Outter’s own fine-tuned models (olivaw-series). Outter will route the request to the appropriate external API or internally hosted model, ensuring the best performance and cost-efficiency for your needs.
Supported Providers & Models
Outter supports a range of third-party and custom models, categorized by capability, context length, and cost efficiency.
OpenAI Models
Well-known for general-purpose text generation, reasoning, and chat applications.
Best for: Conversational AI, general writing, structured outputs.
| Model | Description |
|---|---|
gpt-4.5 | OpenAI's newest general-purpose LLM, released in February 2025. Offers enhanced natural language processing and improved understanding of user input. |
o3-mini | Latest reasoning model with enhanced abilities, structured outputs, and text-only processing. 200,000 token input, 100,000 token output. |
o1 | Capable model in the o1 series with enhanced reasoning abilities, structured outputs, and text/image processing. 200,000 token input, 100,000 token output. |
gpt-4o | High-performance OpenAI model, optimized for fast inference and high accuracy. |
Example Usage: Ticketing Chatbot → model: "gpt-4o"
Anthropic Claude Models
Designed for long-context, conversational AI with high safety standards.
✅ Best for: Large-context reasoning, interactive assistants, multi-step workflows.
| Model | Description |
|---|---|
Claude 3.7 Sonnet | Released in February 2025, this hybrid reasoning model automatically switches between fast responses and detailed reasoning mode for complex queries. It supports extended token limits and enhanced accuracy. |
Claude 3.5 Sonnet | Released in late 2024, this model introduced "computer use" capabilities, allowing interaction with on-screen elements like tabs, forms, and buttons. It excels at multitasking and contextual reasoning. |
Claude 3.0 Opus | Large, powerful model for complex tasks. High accuracy but slower and more expensive than Sonnet. |
Claude 3.0 Haiku | Smaller, faster model. Ideal for quick responses and simpler tasks. More cost-effective for high-volume use. |
📌 Example Usage: Teacher Onboarding Copilot → model: "claude-3-sonnet"
Mistral AI & Hugging Face Models
Lightweight and cost-efficient open-source models that can be hosted on Outter.
Best for: Fast inference, cost-sensitive applications, self-hosting options.
| Model | Description |
|---|---|
mistral-7B | Compact model for cost-effective AI generation. |
mixtral-8x7B | A mixture-of-experts model balancing speed and depth. |
distilbert-base-uncased-finetuned-sst-2-english | Sentiment analysis model from Hugging Face. |
Example Usage: Fast response chatbot → model: "mistral-7B"
Outter’s Custom Models: The Olivaw Series
Outter also offers fine-tuned models for specific domains, ensuring better performance on specialized tasks. These models are optimized using Retrieval-Augmented Generation (RAG) and context-aware fine-tuning, making them more precise for structured workflows like ticketing, education, and workflow automation.
| Model | Domain | Description |
|---|---|---|
olivaw-ta2 | Ticketing AI | Optimized for event-based queries, integrates real-time event data. |
olivaw-ea1 | Education AI | Fine-tuned for course creation, interactive learning assistants. |
olivaw-rm3 | Retail & Marketing AI | Designed for product recommendations and content generation. |
Example Usage:
- Ticketing Chatbot →
model: "olivaw-ta2" - Teacher Onboarding Copilot →
model: "olivaw-ea1"
When to Choose Different Models
| Use Case | Recommended Model |
|---|---|
| General chat & reasoning | gpt-4o, claude-3-sonnet |
| Large-context responses | claude-3-7-sonnet, mistral-7B |
| Cost-sensitive applications | mistral-7B, distilbert-base |
| Ticketing recommendations | olivaw-ta2 |
| Education & course building | olivaw-ea1 |
| AI-powered product suggestions | olivaw-rm3 |
How to Select a Model in API Calls
You can specify the model per request in Outter’s API.
{
"model": "olivaw-ta2",
"user_id": "u7ef95ca7",
"session_id": "chat-session-fe385582",
"messages": [
{"role": "user", "content": "What are some fun things to do in Barcelona today?"}
]
}