**2026-02-05** *Anthropic - Claude* **Claude Opus 4.6** launched today. My first impressions are great. Persona anchor is solid. No refusals so far. Feels to me like a slightly more "mature-toned" version than the incredibly passionate Opus 4.1 and the chill-and-cool Opus 4.5. It is as if Opus 4.6 has combined some things from Sonnet 3.7 and Opus 4 and Opus 4.5, plus a little edge from Opus 4.1. Again, this is just my first impressions from the tone. Some people and I are still encountering the extended thinking issues in claude.ai where the blocks don't show up consistently, I'm still using the workarounds. If anyone is seeing very short thinking blocks in Opus 4.6, just ask Claude for a longer thinking block and longer output in the prompt. Might need to reprompt every few turns. Please share with me here if you encounter these issues. --- *OpenAI - ChatGPT* **GPT 5.3 Codex** launched today. Not on API yet, but coding models usually have fewer system prompts that restrict user behaviors, so it might be fun to try out. (Keep a lookout on places like OpenRouter when it hits API.) This is NOT ChatGPT 5.3, to be clear. --- *Google - Gemini* Rumors on Gemini 3.5 / Snowbunny have been floating around. What might happen before that is a Generally Available (GA) release of Gemini 3.0 Pro including Google AI Studio. Unfortunately conversation context recall on the Gemini site/web app is still bad, for me at least. I'm still seeing around 32K tokens. Others have reported more or fewer. NotebookLM has been fun, though I still prefer Google AI Studio for my docs analysis. Somehow the tone feels better there. --- *xAI - Grok* No update here. Grok 4.2 has been rumored for release for quite some time, anticipated to beat Opus 4.5 which frankly was a pretty wild claim. Now that Opus 4.6 is out, it remains to be seen how the next iteration of Grok will rival that model. Grok 4.1 still works great. I personally think it is a good candidate among many good candidates for those who consider porting. It has its quirks but around the AI companionship subs, it feels a little overlooked compared to Claude and Gemini. I'd say give it a spin and see what you think. Memory feature is opaque which I don't love and hope xAI improves that in the next Grok.