@lightsofapollo2

building gpu cli. any command on a cloud gpu, instantly, across providers. here for agents, inference, and getting a gpu when they're all sold out. boulder

 Boulder, Co
gpu-cli.sh
Joined February 2026
  • Tweets184
  • Following96
  • Followers13
17Media
i haven't started a gpu job by hand in weeks. claude and codex drive them through gpu cli now: grab a gpu across providers, run it, tear it down. great for stateful gpu work gpu-cli.sh

GPU-CLI - Run Code on Remote GPUs

Run code on remote GPUs with a single command. GPU CLI makes remote GPU execution feel like local development.

gpu-cli.sh
2
230
Using review bench to test open code review - not particularly happy with the results with the models I used anyway. I think a better approach might be code review + run the thing + telemetry
1
37
Replying to @ClaudeDevs
@ClaudeDevs using new mods visualize 10 agents over 3 machines with 2 different agent harnesses. Really liking mods so far I wonder if I can push the UI further
1
1
40
`prtl` is acp based cli that allows managing multiple harnesses over different machines. I use tailscale network here for windows, linux and mac
10
Still #1 upgrade to my workflow is @Steve_Yegge's beads. Memory + scaling issues/tasks across machines and team members is essential for more agents in flight
2
33
Staring at thinking traces makes the whole process feel slow. So glad I found this . @opencode wdyt make thinking off a default?
3
26
Getting great use of the -flash models on @ollama . Amazing how capable they have gotten in such a short time.
1
2
39
Been a long time since I've done iOS for real- this is so much nicer. Having astra test, install, clickaround, etc.
22
Replying to @1Password
@1Password moved all my local secrets op cli and it massively helps with worktree setup. I think it would be nice if there was a way to cache select secrets longer so I get prompted less often?
2
2
38
for example development secrets- this is *not* for my production use cases.
2
Liking @opencode 2.0. As a detail item the updater is really good. i.e. update mid session and the ui barely flickered. This is a full update without really disrupting my flow definitely an upgrade over other clis.
1
27
vz 0.4.0 coming soon! think entire dev environments you can spin up in parallel on your Mac. Linux + macOS VMs (ios development!). Private Docker per Linux dev machine. Isolated networks. One project, multiple environments. github.com/gpu-cli/vz
1
2
36
James Lal retweeted
Angus Bezzina
@angus1192Sep 7
Really enjoying working with the new #attn changes that allow me to share thoughts on html files. They're agent agnostic and allows me to quickly provide artefact feedback.
1
2
3
55
Same prompt- 30m more tokens on GLM-5.3 (vs Kimi k3) the results were pretty good but I didn't expect it to run for sooooo much longer
Kimi token usage 1,678,711 vs GLM 5-3 32,323,119

ALT Kimi token usage 1,678,711 vs GLM 5-3 32,323,119

1
2
54
Prompt was to review a series of documents something like: "Review ... entire Storage v2 planning program for architectural correctness, consistency, completeness, and readiness to turn into Beads and implement. ..."
2
22
Had some great early results with Astra but can't help shake the feeling that Fable "gets" what I am trying to achieve better? Not sure why
2
33
underlying security measures for sandboxes (turning kernel flags, etc.) are great for reducing the attack but are not a very good replacement for my full development environment. Feels like true sandbox is one thing and "developer" reproducability is another.
2
26
Going to try @ollama team I think ... Limits on claude + fable feel pretty limiting
1
5
225
Wanted to do a quick proof of concept on kitty graphics protocol now that Ghostty supports it. Seems to work pretty well! github.com/lightsofapollo/br… runnable PoC (this uses electron as the gui backend)

GitHub - lightsofapollo/browser-in-terminal: Stream a real Electron (Chromium) UI into a kitty-gr...

Stream a real Electron (Chromium) UI into a kitty-graphics terminal at 60 fps, with full input forwarding and DevTools rendered in the terminal. - lightsofapollo/browser-in-terminal

github.com
2
28
released 0.15.0 of github.com/gpu-cli/openapi-t… - uses a use validation technique and caught over 3k instances across the 55 specs of cases where we can additionally convert to rich rust types. All fixed.

GitHub - gpu-cli/openapi-to-rust: OpenAPI generator for Rust: typed models, async Reqwest/SSE...

OpenAPI generator for Rust: typed models, async Reqwest/SSE clients, and Axum server scaffolding for OpenAPI 3.0/3.1. - gpu-cli/openapi-to-rust

github.com
3
29
Browser in terminal?
1
11
GUI in terminal (ghostty) 🤔
2
20
James Lal retweeted
GPU CLI@gpucliAug 25
Struggling to build Swift UIs that don't look like slop? We built AnnotKit to help. It's a small tool that annotates your macOS/iOS dev build in place. Just point at a view, type a note, and your agent knows exactly what to change and where. annotkit.gpu-cli.sh

AnnotKit: Native Annotation for AI coding agents

Annotate your macOS or iOS dev build: click a view, type a note, and hand your coding agent a selector it can trace back to code. MIT, Swift 6, pre-1.0.

annotkit.gpu-cli.sh
1
1
1
2
46
Load more