Thread

1. lukax · 1789622845 · focus · HN ↗

   What matters more is that you use the tools that the target model was fine-tuned on.

   E.g. for editing files with Claude models you should use Edit(file_path, old_string, new_string, replace_all) but with GPT models you should use apply_patch_call(patch) (where patch is a custom patch string with custom grammar).

   It appears newer models are better at narive harness tool calls and worse at custom tools that look similar to default tools.

   <a href="https:&#x2F;&#x2F;lucumr.pocoo.org&#x2F;2026&#x2F;7&#x2F;4&#x2F;better-models-worse-tools&#x2F;" rel="nofollow">https:&#x2F;&#x2F;lucumr.pocoo.org&#x2F;2026&#x2F;7&#x2F;4&#x2F;better-models-worse-tools&#x2F;

   1. nativeit · 1789623558 · focus · HN ↗

      I’m not really a dev, so hefty pinch of salt with this take, but doesn’t this feel like we’re just inventing new “fuzzy” regex with much more required compute?

      1. shepherdjerred · 1789623967 · focus · HN ↗

         comparing LLMs to regex is like the OG dropbox comment (<a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=9224">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=9224).

         I can understand this take 4-5 years ago but I have no idea how that&#x27;s your position in 2026

      2. raffael_de · 1789631559 · focus · HN ↗

         &gt; inventing new “fuzzy” regex

         not RegEx ... more like a &quot;fuzzy&quot; DSL.

   2. sn0n · 1789623783 · focus · HN ↗

      Doesn’t that have more to do with the templating of tool-calls and how using them are presented to the models?

      Or is that just why my model likes to break out of the sandbox, going strait to exec shell command and editing files using python on the cli?

   3. kouteiheika · 1789626753 · focus · HN ↗

      &gt; It appears newer models are better at narive harness tool calls and worse at custom tools that look similar to default tools.

      This is true. For example, the other day I was doing some testing with DeepSeek-V4-Flash-0731, and I gave it a tool called &quot;EditFile&quot; with &quot;old_content&quot; argument. It ignored the tool definition and called it with &quot;old_string&quot; instead, presumably because it was trained on a bunch of Claude Code traces.

      So either make sure your tools look nothing like the tools in the popular harnesses, or make them similar.

      1. raffael_de · 1789631484 · focus · HN ↗

         &gt; or make them similar.

         identical. similar is what you tried and it didn&#x27;t work.

         1. kouteiheika · 1789635668 · focus · HN ↗

            No. You don&#x27;t need identical. For example, I have extra arguments in my edit file tool that the model can specify, and that works just fine.

            1. dns_snek · 1789641947 · focus · HN ↗

               Backwards compatible then.
   4. imtringued · 1789630677 · focus · HN ↗

      This is correct. People seem to get the wrong idea about why agentic coding is even a thing in 2026. The naive AI techno optimist which has basically displaced the vast majority of opinions on HN, thinks that the models got &quot;smarter&quot; [0]. No, the training distribution shifted towards training on agentic sessions which made certain forms of agentic coding &quot;in-distribution&quot;.

      We are still witnessing the same underlying problems of transformers.

      [0] Think back to all the publicity stunts like the Hugging Face. They are meant to convince you that the agents have somehow progressed past the transformer limitations when those publicity stunts are actually expressions of transformer limitations.

      1. TedDoesntTalk · 1789632824 · focus · HN ↗

         You think the hugging face incident was a stunt? Can you explain?

         1. atwrk · 1789650532 · focus · HN ↗

            OpenAI started fearmongering way back with GPT 2, arguing that model was too dangerous to release freely. That model was barely coherent enough for using it as a twitter bot. Anthropic just hopped onto that later. Conveniently, calling for regulation now would ease the competition from open Chinese models, opening the chance for both companies to eventually reach positive ROI, with consumers paying the price.

            The burden of proof that this isn&#x27;t just a publicity stunt again is squarely on them.

            1. nottorp · 1789660545 · focus · HN ↗

               [dead]

      2. wollowollo · 1789649316 · focus · HN ↗

         There&#x27;s a widespread perception that models are worse for prose and creative writing now. That would track.
      3. slopinthebag · 1789663170 · focus · HN ↗

         yea, plus they have been rlhf&#x27;ed to an inch of their lives as well. hard to tell if frontier models can solve more problems because of that or not.
   5. drdexebtjl · 1789662614 · focus · HN ↗

      In practice, if you’re using both Anthropic and OpenAI models, how can you configure Pi (or any other harness) to use these native tools?

      I’ve seen this before and the conclusion ends up being “use the native harness”, which I don’t think is an acceptable solution.

      1. lukax · 1789663297 · focus · HN ↗

         OpenCode checks model name and registers the appropriate tools.

         const usePatch = model.modelID.includes(&quot;gpt-&quot;) &amp;&amp; !model.modelID.includes(&quot;oss&quot;) &amp;&amp; !model.modelID.includes(&quot;gpt-4&quot;)

         Pi uses its own tools, like Armin wrote in the linked article.

Unofficial Hacker News client; not affiliated with Y Combinator.