Bug

/model (list models) always returns Error fetching models: OpenAI list models failed: 404 Not Found when using the standard OpenAI config.

Root Cause

In crates/zeph-llm/src/openai.rs:149:

let url = format!("{}/v1/models", self.base_url);

base_url in config.toml is https://api.openai.com/v1 (already contains /v1), so the constructed URL becomes:

https://api.openai.com/v1/v1/models  ← 404

All other endpoints in the same file use the correct pattern without doubling:

• format!("{}/chat/completions", self.base_url) → https://api.openai.com/v1/chat/completions ✓
• format!("{}/embeddings", self.base_url) → https://api.openai.com/v1/embeddings ✓

Fix

Change line 149 from:

let url = format!("{}/v1/models", self.base_url);

to:

let url = format!("{}/models", self.base_url);

Also update the doc comment on line 141 which says GET {base_url}/v1/models``.

Affected

• crates/zeph-llm/src/openai.rs:149
• CompatibleProvider::list_models_remote delegates to OpenAiProvider — also affected
• Reproduces 100% in testing.toml with base_url = "https://api.openai.com/v1"

Severity

Medium — /model command broken for all users with standard base_url containing /v1; chat/embeddings unaffected.