model roundup

Opus 4.6

8 items · started 2026-07-12 · closed 2026-07-19

  1. In february, I claimed the 50$ extra usage Anthropic offered when Opus 4.6 was launched (https://support.claude.com/en/articles/13613973-claude-opus-4-6-extra-usage-promo). I haven't used up all of it (guess my codebases are smaller than y…

  2. Sorry if this is a beginner question. I’m also doing my own research, but I wanted to ask here because I’d really appreciate hearing how people with more experience approach this.

  3. https://preview.redd.it/32bgluogbfdh1.png?width=1083&format=png&auto=webp&s=32620471c7cd6ee6c454e2836aaa2f4f96716f60 # Introducing Factuality in the Arena A new ranking of models is now available based on a **weighted combination of human…

  4. I wrote an askClaude extension for my harness to delegate some reasoning to claude since I use open source models heavily. I was maybe at 30% usage yesterday before I tested my tool (which is just calling claude cli with the appropriate fl…

  5. Six months ago I added a small convention to my CLAUDE.md: sign the work. Every significant file gets a comment block — who made it (me + the model), when, what we were thinking, how confident we were, and what we left unresolved: ```pytho…

  6. Question: How does Opus 4.6 handle the 'Thinking' toggle in the Claude Web App today (Legacy vs. Adaptive)?

  7. Is anyone having issues with Sonnet Subagents at the moment? Opus 4.6 spawns them, they start their task and then freeze.

  8. My personal AI benchmark: porting Super Metroid Map Rando to the SA-1 coprocessor with zero assembly knowledge. Opus 4.6 got it to boot.

← all threads