Understanding Agentforce Vibes limits is essential for optimizing your development workflow. Unlike traditional Salesforce governor limits that apply to Apex execution, Agentforce Vibes has its own set of usage quotas:[11][21]
Daily Request & Token Limits
50
Premium Requests per Org per Day
Uses GPT-5 model
1M
Tokens per Org per Day
Whichever limit is reached first
Unlimited
Overflow Requests
Uses Qwen 3.0 model after quota
Official Quota Definition
From Salesforce Developer Blog: "Each Salesforce org receives a daily quota of 50 premium requests or one million tokens, whichever is consumed first."[11]
Context & File Limits
Agentforce Vibes automatically manages context to stay within processing limits:[21]
- Token limits affect context inclusion - Large files or extensive context may be truncated automatically
- File size restrictions - Very large files may not be fully included in context
- Automatic context management - The system optimizes what gets included to maximize effectiveness
- Each file read consumes a request - Use @mentions to provide context upfront and save requests
Scratch Org Quota Multiplier Strategy
One of the most powerful optimization strategies for extending your daily limits:
How It Works
Every scratch org has its own separate quota of 50 premium requests. This quota is independent of your main org.
The Math
5 scratch orgs = 250 premium requests per day (50 x 5). This dramatically extends your vibe coding capacity on heavy development days.
# Create scratch orgs to multiply your quota
sf org create scratch -f config/project-scratch-def.json -a dev1
sf org create scratch -f config/project-scratch-def.json -a dev2
sf org create scratch -f config/project-scratch-def.json -a dev3
# Each scratch org gives you 50 additional premium requests
# 3 scratch orgs = 150 extra requests per day
Request Conservation Best Practices
- Use @mentions for files - Inject file content directly instead of letting Vibe Codey discover files (each discovery consumes a request)
- Reserve premium for complex tasks - Don't waste premium requests on trivial tasks that can be done via CLI or VS Code extensions
- Use /newchat strategically - Summarize context and start fresh to preserve knowledge without consuming excessive tokens
- Batch related requests - Plan your vibe coding session to maximize value from each request
- Use slash commands - Commands like /deep-planning are optimized for comprehensive output per request
Model Fallback Behavior
When your premium quota is exhausted, Agentforce Vibes automatically falls back to the Qwen 3.0 model:
- GPT-5 (Premium) - First 50 requests or 1M tokens of the day
- Qwen 3.0 (Overflow) - All requests after premium quota is exhausted
- xGen - Salesforce's in-house model, available for certain tasks
Future Paid Plans
Salesforce has announced plans to "launch a way to buy more requests, and the ability to use premium coding models, in a future release." This will allow teams that need more than the free tier to purchase additional capacity.[2]
