[dead]
Cursor Introduces Composer 2.5
171–180 of 238 posts
Re: Cursor Introduces Composer 2.5
#172Say what you want about Cursor but they don’t lack for ambition. Forking VS Code, going big on bleeding edge features like cloud agents, and now they’ve thrown down the gauntlet directly challenging frontier labs by training their own model (“much larger” than Kimi 2.5’s 1T parameters) from scratch. They’ve been highly successful so far. Raised $50B, $2B in revenue, forecast to end 2026 above $6B. But even at these h…
EDIT: As others have pointed out, the comment above contains hallucinations (Like the $50 billion number) and a lot of AI tells. The account doesn’t have a history of AI-like comments but the hallucinations and structure in this one are suspicious. If anything, don’t trust the numbers it cites because they’re made up. Cursor is a team that I want to see succeed. They have stacked their company with very smart people…
Same, I kick the tires on Cursor every several weeks wanting to find they've finally crossed some chasm I can't quite explain. But every time, I bounce off the ground-truth that they're forked off vscode, which just isn't for me. I think moving agents to the center of their experience and developing a model that focuses on speed/efficiency over maximum depth is a promising step away from being a spicy vscode fork.
Re: Cursor Introduces Composer 2.5
#173[flagged]
Re: Cursor Introduces Composer 2.5
#174Earlier quoted context omitted.
Have you tried Zed? I haven’t tried Cursor, so don’t know how they compare, but I like Zed a lot. Anyway, would love to see a comparison from someone who has used a recent version of each.
A few years ago I tried Zed when it was still pretty early, but eventually settled on Cursor. I gave Zed another shot a few days ago because Cursor’s worktree support still feels pretty weak. In my setup I use multiple agents like Claude Code and Codex, and Zed’s ACP support makes it pretty nice to manage them all as “threads” in one place. Worktree switching also feels much smoother. Overall the experience was prett…
Interesting, is it that the tab completion is giving better results, or how it works is better?
Re: Cursor Introduces Composer 2.5
#175Say what you want about Cursor but they don’t lack for ambition. Forking VS Code, going big on bleeding edge features like cloud agents, and now they’ve thrown down the gauntlet directly challenging frontier labs by training their own model (“much larger” than Kimi 2.5’s 1T parameters) from scratch. They’ve been highly successful so far. Raised $50B, $2B in revenue, forecast to end 2026 above $6B. But even at these h…
As a heavy user, I don't think the model is their product. Cursor is primarily a harness and lately, a specialized agent dashboard.
Composer, their in house model, is dispatched by other models like Claude Opus for individual items on a task list. No one is suggesting you write your main prompt to Composer 2.
Re: Cursor Introduces Composer 2.5
#176I have to say the new model is quite good at the basics, I've been handing over more and more tasks from Linear straight to it instead of the copy-paste into Claude dance lately.
At this point, more of my complaints are on the harness side, which is odd since originally they were by far the best harness out there.
Support - This is pretty much non-existant, it's community support or sales support.
Interacting with GitHub - this should work and be awesome, Claude code does this well (responding to lint errors and comments). Cursor you have to poke the agent to look at the comments or lint errors, and even then it's about 10% good. Even GitHub Copilot is better here.
Bugbot - I have it setup to trigger manually, but it still seems to wake up and burn 80-120k tokens just to notice it's configured to be manually invoked. When it does run, it tells me there's no issues (but claude or copilot both find real things)
App - When you have both agent window and the ide windows, it's hard to open up the code in the right directory. A simple "cursor ." from the terminal used to do it, now it'll often open the agent window, you have to try a few times for it to work.
I love that they are running super fast, it's just hard when many of the basics break or don't work.
Re: Cursor Introduces Composer 2.5
#177Say what you want about Cursor but they don’t lack for ambition. Forking VS Code, going big on bleeding edge features like cloud agents, and now they’ve thrown down the gauntlet directly challenging frontier labs by training their own model (“much larger” than Kimi 2.5’s 1T parameters) from scratch. They’ve been highly successful so far. Raised $50B, $2B in revenue, forecast to end 2026 above $6B. But even at these h…
> now they’ve thrown down the gauntlet directly challenging frontier labs by training their own model (“much larger” than Kimi 2.5’s 1T parameters) from scratch.
To clarify, the model Composer 2.5 announced in this post is not that; it uses Kimi 2.5 as a strong starting point. This is not to discount Cursor's work or future ambitions, but one of the most striking things about the last 6 months is that multiple open-source models/labs are now within striking distance of the frontier closed-sourced labs.
See eg Kimi 2.6 benchmarks: https://www.kimi.com/blog/kimi-k2-6
Re: Cursor Introduces Composer 2.5
#178Say what you want about Cursor but they don’t lack for ambition. Forking VS Code, going big on bleeding edge features like cloud agents, and now they’ve thrown down the gauntlet directly challenging frontier labs by training their own model (“much larger” than Kimi 2.5’s 1T parameters) from scratch. They’ve been highly successful so far. Raised $50B, $2B in revenue, forecast to end 2026 above $6B. But even at these h…
Why is this comment upvoted?
It is most likely AI generated with a nice "Raised $50B" hallucination and filled with cliches ("thrown down the gauntlet", "mountain you don’t climb just once", "not for the faint of heart").
Re: Cursor Introduces Composer 2.5
#179Earlier quoted context omitted.
EDIT: As others have pointed out, the comment above contains hallucinations (Like the $50 billion number) and a lot of AI tells. The account doesn’t have a history of AI-like comments but the hallucinations and structure in this one are suspicious. If anything, don’t trust the numbers it cites because they’re made up. Cursor is a team that I want to see succeed. They have stacked their company with very smart people…
Same, I kick the tires on Cursor every several weeks wanting to find they've finally crossed some chasm I can't quite explain. But every time, I bounce off the ground-truth that they're forked off vscode, which just isn't for me. I think moving agents to the center of their experience and developing a model that focuses on speed/efficiency over maximum depth is a promising step away from being a spicy vscode fork.
Cursor 3 is a full rewrite. No VS Code
Re: Cursor Introduces Composer 2.5
#180I have to say the new model is quite good at the basics, I've been handing over more and more tasks from Linear straight to it instead of the copy-paste into Claude dance lately. At this point, more of my complaints are on the harness side, which is odd since originally they were by far the best harness out there. Support - This is pretty much non-existant, it's community support or sales support. Interacting with Gi…
> I've been handing over more and more tasks from Linear straight to it instead of the copy-paste into Claude dance lately
Tangent: we've been using Linear at work and I still don't understand why it claims to be "task tracking for agents". Is there anything at all that lends itself better to agentic workflows compared to JIRA or gitlab/github issues or whatever else?
Seems like Linear just hopped on the buzzword hype train at the exact right moment...