An Empirical Comparision of Claude Pro and ChatGPT Plus (www.reddit.com via reddit)
model roundup
Haiku 4.5
-
Pulled the Artificial Analysis numbers because every thread on this is vibes and no data. Opus 5 beats GPT-5.6 Sol on intelligence, 61 vs 59, which is basically nothing, and Sol does it at half the cost per task ($1.23 vs $2.34).
-
Help me understand my usage cost (www.reddit.com via reddit)
I hit my usage limit for the first time and by using /usage I see this: Session Total cost: $423.97 Total duration (API): 6h 22m 37s Total duration (wall): 4d 22h 34m Total code changes: 4416 lines added, 1163 lines removed Usage by model:…
-
Culling a photo library is a tail-selection problem: you care about the obvious garbage and the standout keepers, not whether photo #412 edges out #487. That shape maps beautifully onto Claude's model tiers, so I built Winnow, an open-sour…
-
Claude keeps refusing to do anything? (www.reddit.com via reddit)
Opus 5/Sonnet 5/Haiku 4.5 After a few messages of us going back and forth working on helping me make an hour by hour schedule for when classes starts in a few weeks to make sure I have time for all of my obligations outside of class this s…