What's this about?
The Fable 5.1 thread reached 962 points and 900 comments — the largest of the day by a wide margin — and its centre of gravity is not the model.
Why does the discussion go to metering rather than capability?
Because capability claims are unfalsifiable on day one and quota is immediately felt. Nobody in the thread can verify 52.6% on Terminal-Bench-Science; everybody can verify that their window reset at an inconvenient time. This is a general property of launch-day discussion and worth knowing when reading it: the loudest signal is the most immediately measurable one, not the most important one.
What's the community saying?
HN 962 points, 900 comments
- "This time it came with a usage reset" is the top comment — the reset, not the model.
- The specific grievance: a reset that pushed a nearly-expired 5-hour window back to a full five hours, costing a cycle in the weekly budget.
- The workaround that circulates: a Haiku cron job to align your windows deliberately.
- Substantive product criticism does appear — alignment checks sensitive enough to push users back to Opus, laziness, and inventing dense conceptual names — from people who had largely given up on Fable.
- One reader notes the irony that a demo about speeding up inference is exactly the thing normal users cannot do with Anthropic's own stack.