Choosing AI Models

Select the right AI model for speed, capability, and credit usage based on your task.

Craft offers several AI models with different strengths. Choosing the right one helps you manage credit usage and match the model to the size and complexity of your task.

Model comparison

Model	Credit usage	Best for	Availability	Underlying model
Max	High	Complex reasoning, multi-document work, long summaries	All platforms (Plus plan and above)	Claude Sonnet 4.5
Fast	Medium	Space-level questions, deeper analysis, longer conversations	All platforms	Claude Haiku 4.5
Core	Low	Everyday in-document queries, balanced performance	All platforms	ChatGPT Nano 5.1
On-device	Free	Short rewrites, quick transformations, simple questions	Mac, iPad, iPhone (document level only)	LLaMa 3.2 1B and 3B + any models available on your device

Credit usage estimates

With 50 credits, you can run approximately:

Model	Approx. queries
Max	~30
Fast	~100
Core	~1,000
On-device	Unlimited (no credits)

These are rough averages from testing. Actual usage varies with prompt complexity, document length, and the amount of context processed.

When to choose each model

On-device

Use for quick edits, sentence rewrites, and simple clarifications. Fast and private. No credit usage.

Best for:

• Fixing grammar or spelling
• Rewording a sentence
• Simple style adjustments
• Quick format changes

Limitations:

• Document-level only (no space queries)
• Limited context window
• Best for shorter content

Core

Use for most general-purpose questions within a document. Balanced speed and capability.

Best for:

• Summarizing a single document
• Answering questions about current content
• Extracting key points
• General writing assistance

Credit efficiency:

• Most cost-effective cloud model
• Good for frequent, everyday tasks

Fast

Use for broad, Space-level queries or when the task requires deeper analysis.

Best for:

• Searching across multiple documents
• Complex space-level queries
• Longer conversations
• Tasks requiring more context

When to use:

• Questions that span multiple documents
• Deeper analytical work
• Multi-turn conversations

Max

Use for long, structured outputs, multi-document synthesis, or anything requiring advanced reasoning.

Best for:

• Research-level summarization
• Complex multi-document analysis
• Advanced reasoning tasks
• High-quality, structured outputs

Requires:

• Plus plan or higher
• Higher credit cost per query

Start with Core for most tasks. Upgrade to Fast for space-level queries. Reserve Max for complex work requiring the highest quality. Use On-device whenever possible for privacy and zero credit cost.

Switching models

You can change models at any time during a conversation:

1. Open the Assistant panel.

2. Click the model name at the top of the panel.

3. Select your preferred model from the dropdown.

Your conversation context is preserved when switching models.

Model availability by platform

Platform	On-device	Core	Fast	Max
macOS	✓ (document)	✓	✓	✓ (Plus+)
iOS	✓ (document)	✓	✓	✓ (Plus+)
Windows	–	✓	✓	✓ (Plus+)
Web	–	✓	✓	✓ (Plus+)

Related Articles

Using Craft Assistant

Credits

Custom Prompts