Upcoming Billing Transition
Agentforce Vibes is introducing changes to access, pricing, and supported LLM models. This topic outlines the rationale, available access models, and transition timeline.
To meet the demands and scale of enterprise-level development, we're shifting to a paid model where we can offer access to higher quality models at scale. This change allows Salesforce to offer increased capacity and better models while managing infrastructure costs.
Starting May 31, 2026, Agentforce Vibes is available through updated access options. Details coming soon.
| Date | Change |
|---|---|
| April 13, 2026 (Done) | Claude Sonnet 4.5 becomes the Pro model, with free limited access updated to 110 requests/1.5M tokens per org per month. GPT-5 mini becomes the fallback model. Previously, the limit was 50 requests/1M tokens per org per day of GPT-5, with GPT-5 mini as fallback. |
| June 1, 2026 |
|
Agentforce Vibes is powered by Large Language Models (LLMs) hosted by Salesforce. The system uses a two-tier model approach: Pro and Core.
Agentforce Vibes automatically starts with the Pro model (Claude 4.5 Sonnet) for chat and interactive experiences. When you reach your monthly Pro model limits, the system automatically falls back to the Core model (GPT-5 mini) for the remainder of the monthly period.
| Feature | Model Used | Usage Limit (per org) |
|---|---|---|
| Agentforce Vibes Chat | Pro. Fallback: Core model. | 110 requests OR 1.5M tokens per month (whichever first) |
| Inline Autocomplete | GPT-4.1 model. | 1,500 requests OR 15,000 tokens per day (whichever first) |
Agentforce Vibes Chat (Pro Model)
- 110 requests per month per org, OR
- 1.5M tokens per month per org
- Whichever limit is reached first
The monthly window resets 30 days after your first Pro model request. When either limit is reached, Agentforce Vibes automatically switches to the Core model.
Agentforce Vibes Chat (Core Model)
- 1500 requests per month per org, OR
- 15M tokens per month per org
- Whichever limit is reached first
Inline Autocompletion (GPT-4.1 Model)
- 1,500 requests per day per org, OR
- 15,000 tokens per day per org
- Whichever limit is reached first
When you reach either limit, no new completions are available until the daily limit resets 24 hours after your first request.