A: Hey Chrissumo12!

Good questions. Let's break it down:
- Grok API ETA: Grok is definitely on our radar and already has a ticket on our public roadmap here: https://agenticflow.featurebase.app/p/please-add-grok-connections. Right now, it hasn't gathered a huge number of votes compared to some other requests, so it's not at the very top of the immediate priority list. The best way to help us bump it up is to head over to that link and cast your vote! We genuinely build based on what the community asks for most.
- Credit Usage for Chat: When you're just chatting back and forth with an agent (like asking simple questions that don't require it to run complex background tasks), the primary credit usage comes from the variable cost – the tokens used by the underlying LLM (like Gemini Flash or your BYOK model).

The fixed cost (4 credits on T1/2, 3 credits on T3/4 per "step") mainly applies when the agent needs to execute a workflow step or use an MCP tool as part of generating its response (for limited time MCP tool call also free). For example, if you ask it to "summarize this webpage," the chat triggers a scraping step and an LLM summarization step, each incurring the fixed cost plus any variable token costs. Simple Q&A usually involves fewer "steps" in the background, so the token cost is the main driver.

Hope that makes sense! Let me know if you have more questions.