Models

Skynet gives you access to frontier models from every major provider through a single workspace. You do not need separate accounts, API keys, or tabs. Every model draws from the same shared memory and unified context.

How Model Selection Works

Skynet offers two ways to interact with models. Both are designed to keep your team focused on output rather than infrastructure.

Smart Routing

By default, Skynet evaluates each prompt and routes it to the model best suited for the task. The system balances quality, speed, and credit cost in real time. A quick summary may go to a lightweight model. A complex reasoning task may go to a frontier model. You do not have to choose. The system optimizes for outcome without spending frontier-model credits on work that does not require them.
Manual Pinning

When you want control, you can pin a specific model to a conversation or project. Pin Claude 3 Opus for deep analysis, or switch to GPT-4o mid-thread without losing context. Your conversation history and memory carry over. You change the engine, not the workspace. Manual pinning is useful when your team prefers a specific model for a known task type, or when a particular model has proven better for a specific output style.

Available Models

The following models are available on the platform. Credit consumption varies by model. Frontier models consume more credits per task. Lightweight models consume fewer.

OpenAI

Model	Modality	Key Capabilities
GPT-4o	Text, Vision, Code	Vision, tool use, function calling, fast multimodal processing
GPT-4o mini	Text, Code	Lightweight and fast, cost-efficient for high-volume tasks
o1-preview	Text, Code	Advanced reasoning, step-by-step problem solving
o1-mini	Text, Code	Reasoning at lower cost, optimized for STEM and logic tasks

Anthropic

Model	Modality	Key Capabilities
Claude 3.5 Sonnet	Text, Vision, Code	Vision, tool use, strong writing and analysis, long context
Claude 3.5 Haiku	Text, Code	Fast and efficient, strong for everyday tasks
Claude 3 Opus	Text, Code	Deep reasoning, complex analysis, long-form generation, long context

Google

Model	Modality	Key Capabilities
Gemini 1.5 Pro	Text, Vision, Code	Vision, very long context window, multimodal understanding
Gemini 1.5 Flash	Text, Code	Fast and lightweight, cost-effective for scaled workloads

DeepSeek

Model	Modality	Key Capabilities
DeepSeek V3	Text, Code	Strong general-purpose performance, cost-efficient
DeepSeek R1	Text, Code	Advanced reasoning, competitive with frontier reasoning models

Meta (via Groq)

Model	Modality	Key Capabilities
Llama 3.1 70B	Text, Code	Open-source flagship, strong general performance
Llama 3.1 8B	Text, Code	Lightweight and fast, ideal for high-throughput tasks

Perplexity

Model	Modality	Key Capabilities
Sonar Large	Text	Web-grounded responses with real-time search
Sonar Online	Text	Lightweight search-augmented model for quick lookups

Mistral

Model	Modality	Key Capabilities
Mistral Large 2	Text, Code	Strong multilingual performance, tool use, efficient inference

Capabilities at a Glance

Not every model supports every capability. Here is how key features map across the catalog.

Vision

Models that can process images alongside text.

GPT-4o
Claude 3.5 Sonnet
Gemini 1.5 Pro

Tool Use and Function Calling

Models that can interact with external tools, APIs, and connectors to execute actions.

GPT-4o
Claude 3.5 Sonnet
Claude 3 Opus
Mistral Large 2
Most flagship models

Long Context

Models that can process large documents, codebases, or extended conversation histories in a single pass.

Gemini 1.5 Pro
Claude 3.5 Sonnet
Claude 3 Opus

Advanced Reasoning

Models built for step-by-step problem solving, mathematical logic, and complex multi-step analysis.

o1-preview
o1-mini
DeepSeek R1

Web-Grounded Search

Models that pull real-time information from the web.

Sonar Large
Sonar Online

Credits and Model Usage

Every model consumes credits when used. The cost depends on the model’s computational requirements.

Lightweight models

GPT-4o mini, Claude 3.5 Haiku, Gemini 1.5 Flash, and Llama 3.1 8B consume fewer credits. They are suited for high-volume, routine tasks where speed and efficiency matter most.
Frontier models

GPT-4o, Claude 3 Opus, Gemini 1.5 Pro, and o1-preview consume more credits. They are suited for tasks that demand the highest quality reasoning, analysis, or multimodal understanding.
Smart Routing

Optimizes credit usage automatically by routing simple tasks to lightweight models and reserving frontier models for complex work.

Because Skynet uses a unified credit pool, you can mix and match providers without managing separate billing relationships. Use Smart Routing for efficiency, and manually pin a frontier model only when the task demands it.

Unified Memory Across Models

When you switch from one model to another mid-conversation, your context carries over. The new model picks up where the last one left off. You do not re-explain your project, re-upload documents, or lose thread history.

This works because the workspace owns the context, not the individual model. Models draw from the same shared memory layer. This makes manual pinning practical: you are continuing with a different engine, not starting a new conversation.

Choosing the Right Model

Scenario	Recommended Approach
Daily tasks and quick drafts	GPT-4o mini, Claude 3.5 Haiku, or Gemini 1.5 Flash
Deep analysis and complex writing	Claude 3 Opus or Claude 3.5 Sonnet
Multimodal tasks with images	GPT-4o, Claude 3.5 Sonnet, or Gemini 1.5 Pro
Long documents and large context	Gemini 1.5 Pro
Step-by-step reasoning and math	o1-preview or DeepSeek R1
Real-time web information	Sonar Large or Sonar Online
Multilingual work	Mistral Large 2
General efficiency	Smart Routing (default)

Get Started

Every model listed here is available in the Skynet workspace. A free tier is available to start testing models across providers.

Try Skynet: Start free and explore models in a unified workspace.

Book a Demo: Schedule a walkthrough to see Smart Routing and model switching in practice.