Hacker News new | past | comments | ask | show | jobs | submit
internet_points | on: Ask HN: What's your current AI agentic setup?
1) What AI subscriptions are you personally paying for? (Claude, ChatGPT/Codex, Cursor, etc)

None.

2) Are you using models outside (or instead) of those subscriptions? For example, pay-as-you-go APIs, OpenRouter, or locally hosted models.

I have a few ChatGPT API credits. I think I've payed at most $40 in one year, using 5.4.

3) What coding harnesses and interfaces are you using? Are you sticking with the default CLIs or IDE integrations, or using alternatives like OpenCode, Pi, Aider, Muse Code, or custom-built harnesses?

Just gptel in a firejailed Emacs. I don't do tool calling. Mostly just code review (have personal scripts set up to get diffs and automatically redact stuff).

4) Are you using any agent orchestration or swarm tools?

No.

5) Are you using broader agent platforms like OpenClaw / Hermes Agent / Grok bot / Dots / Muse?

No.

I'd like to try agent stuff using a local model, but it seems quite expensive to set up. I don't feel comfortable giving so much access to OpenAI/Anthropic.

sigmoid10 | parent
How did you end up with only $40 in API cost despite using AI daily?
internet_points | root | parent
Were you looking at another comment? I had to reread what I wrote, and I can't understand where you'd get the impression that I use it that much. I don't even use it daily (looking at my stats, it seems I used it less than every other day).

gpt5.4 pricing per million tokens is Input: $2.50 Cached input: $0.25 Output: $15.00

I tried re-running a rather long code review request I had lying around, and gptel says it spent 2.2k tokens; I don't know if that's input or output or what, but if we say it's output since that's the most expensive, then that single review cost me just $0.033.