Yeah we have some anti-abuse rules that are over triggering right now, we’re actively working to address this.
@theamolavasare
-
Claude Design transforms rapid prototyping for engineers
By
–
Claude Design is an absolute gamechanger.
— Amol Avasare (@TheAmolAvasare) 17 avril 2026
90% of my scrappy prototyping now happens here. Fastest and most pleasant way for me to convey an idea to eng (with our design systems baked in).@lennysan it took me a ton of self-restraint to not talk about this on the pod 😅 https://t.co/9m7nNwTfqcClaude Design is an absolute gamechanger. 90% of my scrappy prototyping now happens here. Fastest and most pleasant way for me to convey an idea to eng (with our design systems baked in). @lennysan it took me a ton of self-restraint to not talk about this on the pod
-
Opus 4.7 Improves Reasoning With Extended Thinking Tokens
By
–
Opus 4.7 is a much more thorough and precise model. It does use more thinking tokens, and our new tokenizer can also create more tokens (up to 1.35x for the same input content). To offset this, we've increased Opus 4.7 rate limits accordingly on our subscription plans. Enjoy!
-
Web Search Default Settings: No Changes Made
By
–
Responded directly there, this is false. We didn't tweak web search defaults. It's on by default, and we have no intention of changing this.
-
Web Search Default Settings Remain Unchanged by Company
By
–
This is false. We did not tweak the web search default settings. It remains on by default, and we have no intention of changing this.
-
Claude 4.7 Launches xhigh Effort Level for Enhanced Reasoning
By
–
4.7 thinks more, so tokens run a bit higher than 4.6. We launched xhigh, the new default effort level for a finer dial on how hard it thinks vs. how much it spends. Let us know what you think!
-
Opus 4.7 Released: Enhanced Task Handling and Reduced Work Fragmentation
By
–
Opus 4.7 is out! Live on our API, Claude Code, Cowork, and Claude chat. Thing I'm noticing internally: people are re-scoping what they hand to the model. Work that got chunked into small pieces for 4.6 because it was too ambiguous or too long is now going in as one task.