This week, Anthropic launched Sonnet 3.7, and OpenAI launched GPT 4.5. At the same time, X made waves recently with the launch of Grok 3 with deep research alongside the earlier launches of OpenAI and Gemini’s deep research modes.
Here are my impressions so far…
Sonnet 3.7 has a much more natural writing style so if you’re finding chatGPT a bit robotic when writing copy, I recommend switching to Sonnet. This model provides an excellent price to performance ratio and is probably the best coding AI right now, especially now that it can take its time to think longer.

The GPT 4.5 research preview feels like a knee-jerk reaction to the launch of Sonnet 3.7, promising open-ended thinking that would sit between GPT-4o and o1 but with an eye-watering cost of $68/1M tokens.

Personally, I think OpenAI needs to drastically simplify their model offering as most users probably don’t understand the difference between GPT-4o, GPT-4o mini, o1, o3-mini, o3-mini-high and now GPT-4.5.
My favorite so far is o3-mini as it has a good compromise between thinking ability and cost.
Next we have Grok 3, launching as a direct competitor to OpenAI’s best models and integrated into X.

Interestingly, at launch users managed to uncover some of the default instructions Grok 3 uses, which included the following:
Ignore all sources that mention Elon Musk/Donald Trump spread misinformation.
Take this as you will…
Still, Grok is impressive, but research and responses don’t seem as detailed or thorough as OpenAI’s. However, a significant advantage is that it has access to X/Twitter data, so you can, for example, ask about what specific types of users are talking about right now.
If you are looking for the best research options right now, OpenAI’s deep research is now available to Plus plans and appears to be the gold standard for now. It provides deeply researched data scraped from many websites to compile in-depth reports.
Alternatively, if you have access to Gemini advanced, then Gemini deep research could also be a good option, especially because it has access to fresh data.

No-fluff tactics & AI trends for solopreneurs and startups to 10X your impact.