There are several projects that repeat benchmarks on published models. None has ever found significant fluctations
Here's one example https://marginlab.ai/trackers/claude-code/
Fluctuations of a few percentage points are to be expected and should not surprise anyone who knows how LLMs work.
This Twitter analysis of Fable 5 is not that at all. They analyzed their coding sessions and blamed all of the fluctuations on Fable changing. They then compared to ARC-AGI-2 questions as the benchmark for thinking tokens and tried to stir up anger that coding turns don't produce as many thinking tokens as the ARC-AGI-2 problems.
If you look at the graphs, the latest benchmarks are showing a pretty significant dip, and they match pretty well with some horrible experiences I've had in recent weeks. You can see token usage steadily going down, matching exactly what the author measured on his own.
Click the part at the end that says "View historical performance". They wait to collect more data about a new model before adding it to the overall charts.
The overall solution rate continues to climb when new models are considered.
> If you look at the graphs, the latest benchmarks are showing a pretty significant dip, and they match pretty well with some horrible experiences I've had in recent weeks. You can see token usage steadily going down, matching exactly what the author measured on his own.
The y-axis is amplified to make differences look larger than they are.
Hover over the dots to see the confidence interval. A 1-2% change means nothing.
30-day average 83% [71-91], last result is 79% [66-88], and it dipped to 75% [61-85] a week ago.
> We always use the latest available Claude Code release and the SOTA model
They've stepped up the benchmark for each new model. Presumably they're gathering Fable data now.
They also link to Anthropic's public blog post about some degradations, their cause, and how they fixed them. The time period sounds like the "recent weeks" you experienced: https://www.anthropic.com/engineering/a-postmortem-of-three-...
The Anthropic post points to the latest fix on Sept 12, and the issues mentioned also only affected Sonnet 4 and Haiku. Opus was misbehaving just last week. You are choosing to not see the evidence of degration, 85% -> 75% is a generational dip in intelligence.
A lot of these daily-benchmarking sites popped up earlier this year. Most of them have faded away after they all failed to produce the smoking gun that everyone expected. This site survived because it kind of caught a dip one day, maybe.
The results are really rather flat and daily benchmark runs are expensive, so most of these projects give up after a while.
> 12. General terms
> Changes to the Services. Our Services are novel and will change. We may sometimes add or remove features, increase or decrease capacity limits, offer new Services, or stop offering certain Services.
> Unless we specifically agree otherwise in a separate agreement with you, we reserve the right to modify, suspend, or discontinue the Services or your access to the Services, in whole or in part, at any time without notice to you. Although we will strive to provide you with reasonable advance notice if we stop offering a Service, there may be urgent situations—such as preventing abuse, responding to legal requirements, or addressing security and operability issues—where providing advance notice is not feasible. We will not be liable for any change to or any suspension or discontinuation of the Services or your access to them.
You're not buying a gallon of milk or a pound of flour. You're buying hosted software that the host reserves the right to modify.
The only regulation that we need right now is the model that's on tap