model roundup

Haiku 4.5

4 items · started 2026-08-24 · closed 2026-08-31

  1. Subscribe to updates for Degraded performance for Claude Opus 5 and Claude Haiku 4.5 via email and/or text message. You'll receive email notifications when incidents are updated, and text message notifications whenever Claude creates or re…

  2. If you didn't know, Claude's research task is utter dogshit. It goes back to the models that summarize the sources being completely stupid.

  3. Hello everyone, I was wondering if anyone saw an evolution in the ratio of the 5-ho and weekly token allocation. When I started using Claude, late 2025, I was under the impression that filling the whole 5-hour session was filling 10% of th…

  4. evallint evallint audits the reliability of LLM evaluations. Everyone tests their model.

← all threads