Codex mistakenly downloads oss model that already exists locally

What version of Codex is running?

codex-cli 0.80.0

What subscription do you have?

pro

Which model were you using?

qwen2.5-coder:32b

What platform is your computer?

Linux 6.8.0-90-generic x86_64 x86_64

What issue are you seeing?

According to https://docs.ollama.com/integrations/codex, Codex can use local Ollama models with its --oss <model_name> flag. However, if I specify a model that I've already downloaded to my localhost using ollama pull <name>, Codex attempts to re-download it. As these models can be quite large, this is a huge waste of resources.

What steps can reproduce the bug?

1. Download a model using ollama. e.g. ollama pull qwen2.5-coder:32b
2. Run codex with this model, e.g. codex --oss qwen2.5-coder:32b.
3. Notice codex tries to redownload the model, e.g.

localhost:~$ ollama list
NAME ID SIZE MODIFIED
codellama:34b 685be00e1532 19 GB 2 hours ago
deepseek-coder-v2:16b 63fb193b3a9b 8.9 GB 2 hours ago
deepseek-coder:33b acec7c0b0fd9 18 GB 2 hours ago
qwen3-coder:30b 06c1097efce0 18 GB 4 hours ago
qwen2.5-coder:32b b92d6a0bd47e 19 GB 6 hours ago
qwen3:14b 7d7da67570e2 9.3 GB 8 months ago
localhost:~$ codex --oss qwen2.5-coder:32b
Downloading model: total 12.85 GB
0.38/12.85 GB (2.9%) 104.7 MB/s

What is the expected behavior?

If the ollama model is listed under ollama list then codex shouldn't redownload the model.

Additional information

_No response_