Papers, benchmarks, lab announcements.
A cluster of recent papers points the same direction: a smaller model that checks its own work beats a bigger model that doesn't. The economics of inference are quietly being rewritten.
2 min read
