‹ BackHN Continuity

Thread

Getting the most out of Opus 5.5 in Claude and Claude Code

232 points · 156 comments · saikatsg

  1. rdli · · focus · HN ↗
    It’s a really good model. Over the past few days, I give Opus some general directives to basically speed up our CI, and telling it I care both about billing minutes and wall clock time. I told it to create a plan after analyzing everything in our CI, run the plan by a Fable subagent, and then focus on low-risk, high-reward changes.

    9 hours later, I had 12 PRs ready to be merged, and the net result is CI time has dropped from ~10 minutes to ~4 minutes, and billing minutes have dropped around 60%. Less than an hour of my attention.

    1. chewchewchew · · focus · HN ↗
      9 hours?!
      1. rdli · · focus · HN ↗
        Yes. It spawned multiple subagents to run different experiments to benchmark a lot of different things, reviewed CI logs from past runs, etc. In the end, there were changes to what/how we cached, various code quality checks, speeding up test runners, and many other things.
        1. Tade0 · · focus · HN ↗
          I dare not ask about the cost, having burned $60 on a task running for 1h 16min once.
          1. rdli · · focus · HN ↗
            I’m on the $100/month subscription; this session took about $500 in token-equivalent costs.

            (Note that it wasn’t all Opus 5.5; I have a setup that uses Fable 5.1 as an advisor, Sonnet 5.5 for mechanical changes, etc.)

            1. tripleee · · focus · HN ↗
              God I hope the prices drop quick. Once they stop subsidizing it these kinds of workflows will be unaffordable for anyone who isn't already wealthy
              1. satvikpendem · · focus · HN ↗
                They have dropped, in Chinese models.
              2. wyre · · focus · HN ↗
                Pricing is dropping quick. Inference is so cheap, I think they are losing a lot less money selling subscriptions than you think. It might even be more expensive managing the load, than actually selling the tokens at subscription prices.

                We are seeing with OpenAI, allegedly through their new pricing scheme, as intelligence and model efficiency increases they offer the same throughput while advertising 1/2 as much usage, letting Astra consume more usage, essentially only being available to those wealthy enough to afford it while still offering essentially unlimited Sol and Luna to their subscription tiers.

                Also if you're cache hit rate is high enough a billion tokens tokens from Deepseek 4.1 Flash costs less than $15.

              3. debatem1 · · focus · HN ↗
                Assuming this isn't some toy CI a 60% drop in billable minutes will make $500 back pretty quick. Github is wildly expensive.
                1. simon-b · · focus · HN ↗
                  The cost of the standard `actions_linux` is $0.006/minute, so spending $500 to save 6m per invocation, break-even is at ~14k invocations. But, if they're using larger machines and/or parallel jobs the $$$ saving accrues faster. IMO the wall-time saving shortening feedback loop may be a bigger win, but harder to value.
              4. miroljub · · focus · HN ↗
                Subsidies are a lie planted by Misanthropic and ClosedAI to milk their users and let them think it's the other way round.

                Inference is highly profitable business, even for third parties with much less resources and expertise.

                1. azan_ · · focus · HN ↗
                  Inference is highly profitable, but you need to recoup losses from training.
                  1. miroljub · · focus · HN ↗
                    Only if your training costs are overblown because you are bruteforcing it. Otherwise it's an equivalent of printing money.
              5. Sevii · · focus · HN ↗
                An insane amount of compute manufacturing comes online in 2028. Compute is a commodity. It's not going to stay expensive for long.
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.