Skip to content

Drop the Ollama handling - #98

Merged
CrypticSwarm merged 3 commits into
masterfrom
ollama-tong
Oct 11, 2026
Merged

CrypticSwarm merged 3 commits into
masterfrom
ollama-tong

Conversation

@CrypticSwarm

Copy link
Copy Markdown
Owner

Ollama moves to a shared tong in its own repo, built on tong GPU access (resources.gpus) and named volumes. This removes what it replaces:

  • the run_ollama, logs_ollama, stop_ollama, and run_<model> targets, the OLLAMA_* variables, and the ollama/ store
  • the ollama provider in opencode/opencode.json
  • the Ollama mentions in AGENTS.md and the docs; the GPU model-server tong example is now generic

gpu_stat stays. Tests that use ollama only as a sample name are unchanged.

A checkout that still has an ollama/ store will now show it as untracked; move it into the tong's volume or delete it.

Ollama now ships as a shared tong in its own repo, using tong GPU
access and a named volume, so the run/stop/logs/model targets, the
OLLAMA_* knobs, and the in-checkout ollama/ store go.
With no Ollama container in this repo the provider points nowhere. The
tong's repo carries the block for anyone who enables it.
@CrypticSwarm
CrypticSwarm merged commit f97b231 into master Oct 11, 2026
4 checks passed
@CrypticSwarm
CrypticSwarm deleted the ollama-tong branch October 11, 2026 19:33
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant