nerfWatch()
Daily tests and community votes on whether AI models got nerfed — each model compared only with its own first week.
Last verified:
What is nerfWatch()?
nerfWatch() is a daily monitoring service that tests whether AI models (e.g., Claude, Grok, GPT, Meta models via OpenRouter) have been 'nerfed' or degraded over time. Each model takes the same benchmark test daily and is compared only against its own first-week performance, then users vote on perceived quality changes.
nerfWatch() pricing
Pricing model: Freemium
Free
nerfWatch() pros
- Daily automated benchmark testing of models against their own baseline
- Community voting lets users compare subjective feelings with objective numbers
- Email alerts when a model's nerf verdict changes
- Fair comparison methodology — models only compared to their own first week
nerfWatch() cons
- Current data is sparse, with most models marked 'too soon' or 'too few votes'
- Limited model coverage shown on the page
- Relies partly on subjective community votes
Frequently asked questions about nerfWatch()
How often are models tested?
Every model takes the same test every day, compared only with its own first week of results.
What do I get if I subscribe?
One email when a model's verdict changes — nothing else.