First off, ChatGPT is still dominating with around 5.5 billion visits, but the growth has slowed to just 6.82%. Meanwhile, Gemini is going crazy with 28.9% growth - that's huge. Google pushed hard on the multimodal thing, and apparently it's working. I'm seeing more people switch to Gemini for projects that need image handling, though I haven't personally tested it in production yet.
What really caught my attention is DeepSeek R1. It's sitting at rank 4 on the global list and it's actually got some decent traction. China's been doing interesting work in the reasoning model space, and this one's reportedly faster than older versions - maybe 3x improvement according to some benchmarks I stumbled across. The Chinese market is obviously huge, but whether it scales internationally is something I'm curious about. No idea if it'll stick for long-term use.
And oh, Claude 3.7 Sonnet - the programming scores are wild at 91.2 on HumanEval. That number matters if you're actually shipping code. I've seen developers specifically choose it for longer documents (100k tokens is no joke), and the safety compliance side is stronger than some alternatives. Not perfect, but solid.
The honest take? We're definitely past the "try everything" phase. People are settling into their preferred tools now. The ones with real improvements in speed and specific use cases are the ones still growing. The rest are stagnating - there's your signal right there.
One thing though - these numbers shift monthly. Some rankings from March might look totally different now. Worth checking back if you're tracking this stuff for work.
Top comments (0)