Commentary on X
@vipulved @cursor_ai Congrats Vipul and the Together team, powering Cursor at this scale is impressive. The custom kernels and orchestration layer sound like the real differentiator. What’s been the biggest surprise in optimizing for their real-time editor workloads?
@togethercompute @cursor_ai Strong inference infra win. Curious what the latency/per-token cost spread looks like vs. their previous setup — those benchmarks would tell the real scaling story.
@togethercompute @cursor_ai That's awesome, congrats to the whole team. Must feel great seeing it all come together in production. Keep crushing it.