nytoast I have an ultimate subscription. Yesterday I noticed that I couldn't pick Sonnet with reasoning. Any ideas as to why? I expected it to work as usual and allow me to choose between regular and reasoning models for various tasks.
mb I believe the reasoning and non-reasoning models have been combined and you can now (since 10 minutes ago) select an effort level for the model instead. Is there any reasoning model you can't select the effort for?
source & further reading
kagifeedback.org — original article
Non-ASCII characters in q= are not decoded when Assistant is opened via a URL template (umlaut stays as %C3%BC)
Add keyboard shortcuts to switch models
Assistant artifacts (a la Anthropic Claude)