Newbie set ups #706

Closed
opened 2026-02-16 17:27:58 -05:00 by yindo · 3 comments
Owner

Originally created by @mrbm on GitHub (Jul 14, 2025).

Originally assigned to: @thdxr on GitHub.

Hey guys, I want to vibe code so bad...want to see what this magic is.

I have been messing around with this project, trying to get it do something pretty basic in the vibe world: create a slightly more than basic react app that uses mapbox. I've got open code running on my mabook pro m4 24 gig computer, and then i've got ollama running on the same network on another computer (m4 mini 24 gig). I've tried the Qwen 3 model 32B quant with a 16k context window, but seems to just eventually freeze and not doing anything besides mapping out the project. I've tried the Qwen 2.5 with a 16k context window, can't seem to use the tooling. I've also tried the openrouter free models, but again, it starts, seems to be doing some stuff but then when it gets to writing it kinda just stops.

I've tried reaching the ollama via a cloudflared tunnel, as well as on a local ip when on the same network. Either way, getting freezes in the project before anything gets done and just can't seem to do the job. Qwen 3 came the closest to writing some stuff (nothing fully working) and 2.5 wrote stuff (went halfway) to the screen (but highlighting stuff on the screen for some reason is blocked) so both are a fail.

I want to experience the VIBE!

For the people that have something working:

  1. Can you tell me how you can copy and paste stuff from the screen of the opencode terminal?
  2. Can you tell me your set up of open router and what model you run on a project that is useable.
  3. Are you doing it from the terminal or do you have it hooked up to the editor
  4. Have you created a react project from nothing?
  5. What do you think are the min requirements on a mac for ollama and would be the model for something that is useful to experiencing vibe.

I think this project is the future, but I have yet to experience the vibe...love to hear from some people.

Originally created by @mrbm on GitHub (Jul 14, 2025). Originally assigned to: @thdxr on GitHub. Hey guys, I want to vibe code so bad...want to see what this magic is. I have been messing around with this project, trying to get it do something pretty basic in the vibe world: create a slightly more than basic react app that uses mapbox. I've got open code running on my mabook pro m4 24 gig computer, and then i've got ollama running on the same network on another computer (m4 mini 24 gig). I've tried the Qwen 3 model 32B quant with a 16k context window, but seems to just eventually freeze and not doing anything besides mapping out the project. I've tried the Qwen 2.5 with a 16k context window, can't seem to use the tooling. I've also tried the openrouter free models, but again, it starts, seems to be doing some stuff but then when it gets to writing it kinda just stops. I've tried reaching the ollama via a cloudflared tunnel, as well as on a local ip when on the same network. Either way, getting freezes in the project before anything gets done and just can't seem to do the job. Qwen 3 came the closest to writing some stuff (nothing fully working) and 2.5 wrote stuff (went halfway) to the screen (but highlighting stuff on the screen for some reason is blocked) so both are a fail. I want to experience the VIBE! For the people that have something working: 1. Can you tell me how you can copy and paste stuff from the screen of the opencode terminal? 2. Can you tell me your set up of open router and what model you run on a project that is useable. 3. Are you doing it from the terminal or do you have it hooked up to the editor 4. Have you created a react project from nothing? 5. What do you think are the min requirements on a mac for ollama and would be the model for something that is useful to experiencing vibe. I think this project is the future, but I have yet to experience the vibe...love to hear from some people.
yindo added the model-problem label 2026-02-16 17:27:58 -05:00
yindo closed this issue 2026-02-16 17:27:58 -05:00
Author
Owner

@thdxr commented on GitHub (Jul 16, 2025):

these are all limitations with non- high end models

cheap/local models just cannot handle a workload like this yet

@thdxr commented on GitHub (Jul 16, 2025): these are all limitations with non- high end models<br><br>cheap/local models just cannot handle a workload like this yet
Author
Owner

@thdxr commented on GitHub (Jul 16, 2025):

closing issue for now

@thdxr commented on GitHub (Jul 16, 2025): closing issue for now
Author
Owner

@mrbm commented on GitHub (Jul 16, 2025):

hey dax, would it be possible to suggest any min set up at all? Addiitonally I have had it making some stuff, the bigger issue is just opencode timing out for no explaination. If it didn't time out, i think lower end systems like a mac mini m4 24 gig solely serving ollama with Qwen3:14B quant with a 65k context window could get the job done. I encourage you to review this a bit more.

@mrbm commented on GitHub (Jul 16, 2025): hey dax, would it be possible to suggest any min set up at all? Addiitonally I have had it making some stuff, the bigger issue is just opencode timing out for no explaination. If it didn't time out, i think lower end systems like a mac mini m4 24 gig solely serving ollama with Qwen3:14B quant with a 65k context window could get the job done. I encourage you to review this a bit more.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: anomalyco/opencode#706