Docs/Models & Variants

Models & Variants

ModelHub provides access to 6 AI personalities, each with multiple variants optimized for different use cases. Understanding their strengths helps you choose the right model for each task.

Choosing a Model

Each model has a unique personality and approach. Some are better suited for analytical work, while others excel at creative tasks or teaching. Experiment to find which works best for your needs.

Tip: You can switch models mid-conversation. Try different models for different parts of your workflow — use one for brainstorming and another for implementation.

Understanding Variants

Each provider offers multiple variants with different characteristics:

Speed

Response time: Fast, Balanced, Methodical

Reasoning

Depth: Basic to Exceptional

Context

Window: 8K to 1M+ tokens

Cost

Tier: $ to $$$$

Available Providers

Claude

Claude

Thoughtful, careful, long-form reasoning

Available variants:

Opus-class
Methodical1M+$$$$
Sonnet-classDefault
Balanced200K$$$
Haiku-class
Fast128K$$
Instant-class
Fast32K$
GPT

GPT

Versatile, capable, widely knowledgeable

Available variants:

GPT-5.2
Balanced1M+$$$$
GPT-5.1Default
Balanced200K$$$
GPT-5
Balanced200K$$$
GPT-4.5
Fast128K$$
GPT-4.1
Fast128K$$
GPT-4.1-mini
Fast32K$
Grok

Grok

Direct, witty, unfiltered insights

Available variants:

Grok-3
Balanced200K$$$$
Grok-2Default
Balanced128K$$$
Grok-2-mini
Fast32K$$
Grok-Fast
Fast32K$
Grok-Deep
Methodical1M+$$$
Gemini

Gemini

Analytical, multi-modal, research-oriented

Available variants:

Ultra-class
Methodical1M+$$$$
2.0 ProDefault
Balanced200K$$$
2.0 Flash
Fast128K$$
1.5 Pro
Balanced200K$$
1.5 Flash
Fast128K$
Nano
Fast8K$
Kimi

Kimi

Friendly, patient, educational focus

Available variants:

Kimi-Reasoner
Methodical200K$$$
Kimi-ProDefault
Balanced128K$$
Kimi-Standard
Balanced128K$$
Kimi-Mini
Fast32K$
Kimi-Coder
Balanced128K$$
Meta AI

Meta AI

Open, collaborative, community-focused

Available variants:

Llama 4
Balanced200K$$$
Llama 3.3Default
Balanced128K$$
Llama 3.2
Balanced128K$$
Llama 3.1
Fast32K$
Llama 3.1 Small
Fast32K$
CodeLlama
Balanced128K$$

Recommendations by Task

Code Generation

GPT (GPT-5.1, GPT-5.2) or Claude (Sonnet-class) — fast, accurate, follows patterns well

Complex Analysis

Claude (Opus-class) or Gemini (Ultra-class) — deep reasoning, thorough explanations

Learning & Tutorials

Kimi (Kimi-Pro) — patient, educational, step-by-step approach

Quick Tasks

Any "mini" or "flash" variant — fast responses, cost-effective

Related