Ask HN: How long before agentic IDEs/CLIs cut you off from pro-grade models?

Many agentic IDEs and CLIs provide free or paid access to pro-grade models such as Claude Sonnet/Opus, Gemini 2.5 Pro, and GPT o3.

Vendors typically describe limits using tokens, credits, or requests, but that doesn’t always match the actual coding hours before hitting limits. I'd love to hear your real-world experiences.

Reply using this format:

[Tool Name] [Tier] [Usage Type] – [usage per day/week/month]

Usage Types:

[a] Heavy: Intensive coding, vibe coding, major refactors.

[b] Medium: Regular coding, debugging, smaller refactors.

[c] Low: Occasional quick fixes, snippets, reviews.

If you haven't hit limits yet, put NL (no limit).