Ask HN: How do GitHub Copilot and Anthropic Claude compare in cost? | Hacker News Reader

Ask HN: How do GitHub Copilot and Anthropic Claude compare in cost?

4 pointserwindegroot7 comments

At the company I work at the developers all have a GitHub Copilot subscription. Copilot allows them to use several models, including Claude. From the usage I can actually see that Claude Sonnet and Opus are the most commonly used models. This got me thinking: would it be more efficient to switch to Claude completely? It is difficult to compare the GitHub credits with the Claude tokens, but does someone have experience using both? Is the difference significant?

View on HN

Claude is the most expensive route for enterprise. With copilot at least you can choose cheaper models if things get too expensive and restrict them in the worst case. None are cheap though you’re talking hundreds of thousands to millions of dollars a year if you have over a thousand people using it.

Anecdotally, you will get more mileage out of a Claude or ChatGPT plan than with a GitHub Copilot plan.

If you're worried about costs, you might try using subagents with a cheaper model or an open model like Kimi/GLM and use Claude/Codex as the orchestrator. This is very efficient.

Err why dont you ask them both and compare their answers ?

That's a good suggestion. I actually asked Copilot, but the answer seems to biased. It wouldn't give a clear answer on how the two compare, but it suggested that I use both side by side (copilot for inside the IDE and Claude for 'extra' prompts).

I was hoping someone had some experience with this. I would definitely value the conclusion of an actual human on this :)

So you dont trust them, but you want to trust them ?

Don't be cheeky. I never said anything about trust. And as a rule you should never completely trust AI and always check the results. That doesn't mean you can't use AI.

Do you know what the major problem with AI generated code is ? Have you ever done a code review with lots of files ?

The larger the text to review the less anyone can comprehend and actually critically understand and comment.

Look back at your history of reviews, which ones actually get more comments and which just get pushed thru.. ?

Reply on news.ycombinator.com