"Model metadata for `gpt-oss:20b` not found" with ollama

[loonery]

What version of Codex CLI is running?

0.114.0

What subscription do you have?

None - I am using an Ollama provider

Which model were you using?

gpt-oss:20b

What platform is your computer?

Linux 6.18.17 x86_64 unknown (NixOS)

What terminal emulator and version are you using (if applicable)?

wezterm

What issue are you seeing?

When I try to interact with the Codex CLI, I see the warning message in the attached screenshot and quoted below. This was not a problem before using the 0.114.0 version of codex cli.

⚠ Model metadata for gpt-oss:20b not found. Defaulting to fallback metadata; this
can degrade performance and cause issues.

As the message notes, performance is degraded severely and it appears that when I try to interact with the agent after the message, that features like "plan mode" are not usable and the agent completely loses track of its goal.

Relevant config:

model = "gpt-oss:20b"
model_provider = "ollama"

[model_providers.ollama]
baseURL = "http://localhost:11434/v1"
name = "Ollama"

I have not found a workaround. I have found seemingly related issues (see list below), but none of the workarounds discussed there are working for me.

• Model metadata for gpt-5-codex not found. Defaulting to fallback metadata; this can degrade performance and cause issues #12100 (I have tried feeding a relevant model_catalog_json path to the config)
• ollama launch codex - ⚠ Model metadata for gpt-oss:20b not found. Defaulting to fallback metadata; this can degrade performance and cause issues. ollama/ollama#14752 (duplicate issue on the ollama repo)
• find_model_by_namespaced_suffix rejects hyphens in namespace, breaking custom provider slugs like llm-gateway/gpt-5.4 #14276 (I have tried using a model with an un-hyphenated name)

What steps can reproduce the bug?

Open codex cli v0.114.0 configured to use gpt-oss:20b with ollama 0.17.7. Chat with the agent and see the message.

What is the expected behavior?

No warning message, agent stays on task.

Additional information

No response

[github-actions]

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

• find_model_by_namespaced_suffix rejects hyphens in namespace, breaking custom provider slugs like llm-gateway/gpt-5.4 #14276

Powered by Codex Action

[loonery]

I have mentioned the above potential duplicate in my issue. I do not think that this is a duplicate issue.

[RoterLong]

May I ask if it has been solved？

[loonery]

@RoterLong - it has not been solved.

In addition to the description that I left of the "missing metadata" I've also realized that the performance degradation I'm noticing is this bug: #11940 where the gpt-oss model no longer has access to 'apply_patch'

[ninevoltz]

model-catalog.local.json

You need a "model-catalog.local.json" in your "~/.codex" directory.
Include it inside your config.toml with: model_catalog_json = "~/.codex/model-catalog.local.json"

To be honest, it feels like using local models with codex is going to end at some point. I'm amazed they included it at all, since it doesn't benefit OpenAI financially.

codex/codex-rs/core/models.json in the source tree has examples.

[loonery]

@ninevoltz thank you for the response. That is working for me now. Originally it didn't work, but that was because the "slug" and "display_name" keys were "gpt-oss-20b" which did not match my specified model of "gpt-oss:20b".

[loonery]

Closed this issue due to the discovered workaround.