Andy Halliday

So I'll have three primary subscriptions.

I'll have Anthropic, I'll have Gemini, and I'll have Perplexity, which I use in a different way, but is increasingly capable of doing almost all the same things that Gemini and or Anthropic can do.

with some additions that are beyond and distinct from those two.

And then I also will maintain a subscription to Lovable because I like that platform and its ease of use.

So that's just my response to what's happening out there in this confrontation, competitive confrontation between OpenAI and Anthropic.

On the subject of coding, I think it's notable that Cursor just released Composer 2.

Yeah.

And Composer 2 beats Opus 4.6.

I'm not sure how well it compares to Codex.

I think Codex is at the top of the leaderboards right at the moment.

With the advent of GPT 5.4 as the reasoning engine behind Codex or some of the...

benefits of the training of GPT-5.4 probably

are leaked over to Codecs in that way.

But Cursor Composer, it takes advantage of multiple models.

You can use multiple models with it, but it's an orchestration or a tool framing that advances the capability of Cursor beyond what just a straight Opus 4.6 can do with coding, for example.

On benchmarks.

And I got to believe that if Cursor, with its very large group of users, is training on coding specifically in a way that Anthropic or OpenAI don't train just on coding...

They also are working heavily on training for general reasoning models.

But if Cursor is really achieving that level of benchmark performance, I got to believe that the people using Cursor 2 now probably have a step above what's available through these other coding platforms.

Okay.