fix(llm): suppress thinking mode for qwen3/gemma4 on Ollama /v1 endpoint (#3228)

* fix(llm): suppress thinking for qwen3/gemma4 on Ollama /v1 compat endpoint

When using qwen3, QwQ, gemma4, or other thinking models via Ollama's
OpenAI-compatible /v1 endpoint, the model routes all output into its
<think>...</think> reasoning block. Since Odysseus strips thinking
content from round_response and only accumulates native tool_calls,
this produces a round with 0 chars, 0 native calls, 0 tool blocks —
the agent appears to silently do nothing.

Root cause: Odysseus classifies the /v1 endpoint as provider="openai"
(not "ollama"), so the payload is built as a standard OpenAI payload
without any Ollama-specific options. Ollama's /v1 endpoint accepts
"think": false as a top-level parameter to suppress extended thinking,
but this was never sent.

Fix:
- Add _is_ollama_openai_compat_url() to detect local Ollama /v1 URLs
- Inject "think": false in both stream_llm and llm_call_async for
  thinking models (qwen3, QwQ, gemma4, DeepSeek-R1, etc.) on this
  endpoint

Verified with qwen3:14b on Ollama 0.24: with think=False the model
correctly emits native tool_calls in a single streaming chunk and
the agent executes bash/file/web tools as expected.

* fix(llm): extend _is_ollama_openai_compat_url to match localhost on any port

Per reviewer feedback on PR #3228:

1. Generalize host detection to mirror _is_ollama_native_url: match any
   localhost/127.0.0.1/0.0.0.0/::1 host (not just port 11434) so that
   custom OLLAMA_HOST ports and container remaps are also covered.

2. Add tests/test_llm_core_ollama_thinking.py covering:
   - _is_ollama_openai_compat_url for all positive/negative URL cases
     including IPv6, non-default port, native /api path, and real OpenAI
   - Payload injection: think:false set for Ollama /v1 thinking model,
     not set for non-thinking model, not set for real OpenAI endpoint,
     and set for localhost on a non-default port (the new case)

src/llm_core.py

@@ -276,6 +276,24 @@ def _is_ollama_native_url(url: str) -> bool:
     return local_ollama_host and (path == "" or path == "/api" or path.startswith("/api/"))
 
 
+def _is_ollama_openai_compat_url(url: str) -> bool:
+    """Return True for local Ollama's OpenAI-compatible /v1 surface.
+
+    Mirrors the host detection used by ``_is_ollama_native_url`` so that the
+    two helpers stay in lockstep: a localhost Ollama on a non-default port
+    (custom ``OLLAMA_HOST``, reverse proxy, container port remap) is treated
+    the same way here as it is on the native ``/api`` path.
+    """
+    try:
+        parsed = urlparse(url or "")
+    except Exception:
+        return False
+    host = parsed.hostname or ""
+    path = (parsed.path or "").rstrip("/")
+    local_ollama_host = host in {"localhost", "127.0.0.1", "0.0.0.0", "::1"} or parsed.port == 11434
+    return local_ollama_host and (path == "/v1" or path.startswith("/v1/"))
+
+
 def _ollama_api_root(url: str) -> str:
     """Return a native Ollama API root such as https://ollama.com/api."""
     url = (url or "").strip().rstrip("/")
@@ -1344,6 +1362,9 @@ async def llm_call_async(
         if max_tokens and max_tokens > 0:
             tok_key = "max_completion_tokens" if _uses_max_completion_tokens(model) else "max_tokens"
             payload[tok_key] = max_tokens
+        # Suppress thinking for qwen3/gemma4 on Ollama /v1 — same as stream_llm.
+        if _is_ollama_openai_compat_url(url) and _supports_thinking(model):
+            payload["think"] = False
 
     if _is_host_dead(target_url):
         raise HTTPException(503, f"Upstream {_host_key(target_url)} marked unreachable (cooldown active)")
@@ -1461,6 +1482,11 @@ async def stream_llm(url: str, model: str, messages: List[Dict], temperature: fl
             payload[tok_key] = max_tokens
         if tools:
             payload["tools"] = tools
+        # For Ollama's OpenAI-compat /v1 endpoint with thinking models (qwen3,
+        # gemma4, etc.), suppress thinking so tool calls aren't swallowed inside
+        # <think> blocks. Ollama /v1 accepts "think": false as a top-level param.
+        if _is_ollama_openai_compat_url(url) and _supports_thinking(model):
+            payload["think"] = False
         h = _provider_headers(provider, headers)
         if provider == "copilot":
             from src.copilot import apply_request_headers