‹ BackHN Continuity

Thread

Best LLM for every budget, updated daily

184 points · 113 comments · terryds

  1. ford · · focus · HN ↗
    You're better off going directly to artificial analysis, this is a feature-poor/misleading/outdated repackaging

    Ex. this type of price estimation is quite naive - some models can require 2-3x the number of tokens to achieve the same level of intelligence. Artificial Analysis' own cost per task is a more fair estimation of cost.

    1. pickledish · · focus · HN ↗
      Agree :) direct link for those curious:

      <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models#price-cost" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models#price-cost

      <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models?models=gpt-6-luna%2Cglm-5-3-flash%2Cmuse-spark-1-3%2Cqwen3-8-flash-next%2Cmimo-v2-6-pro#cost-tabs" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models?models=gpt-6-luna%2Cglm...

      1. neuronexmachina · · focus · HN ↗
        It&#x27;s kind of crazy that claude-opus-5.5 (max) uses 119k tokens per &quot;intelligence index task&quot; while gpt-6-astra (max) uses 27k.

        <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models?cost=intelligence-vs-cost-per-task#token-use" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models?cost=intelligence-vs-co...

Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.