language_models: Add thinking support for ollama (#31665)

This PR updates how we handle Ollama responses, leveraging the new
[v0.9.0](https://github.com/ollama/ollama/releases/tag/v0.9.0) release.
Previously, thinking text was embedded within the model's main content,
leading to it appearing directly in the agent's response. Now, thinking
content is provided as a separate parameter, allowing us to display it
correctly within the agent panel, similar to other providers. I have
tested this with qwen3:8b and works nicely. ~~We can release this once
the ollama is release is stable.~~ It's released now as stable.

<img width="433" alt="image"
src="https://github.com/user-attachments/assets/2983ef06-6679-4033-82c2-231ea9cd6434"
/>


Release Notes:

- Add thinking support for ollama

---------

Co-authored-by: Bennet Bo Fenner <bennetbo@gmx.de>

crates/ollama/src/ollama.rs

@@ -38,6 +38,7 @@ pub struct Model {
     pub max_tokens: usize,
     pub keep_alive: Option<KeepAlive>,
     pub supports_tools: Option<bool>,
+    pub supports_thinking: Option<bool>,
 }
 
 fn get_max_tokens(name: &str) -> usize {
@@ -67,6 +68,7 @@ impl Model {
         display_name: Option<&str>,
         max_tokens: Option<usize>,
         supports_tools: Option<bool>,
+        supports_thinking: Option<bool>,
     ) -> Self {
         Self {
             name: name.to_owned(),
@@ -76,6 +78,7 @@ impl Model {
             max_tokens: max_tokens.unwrap_or_else(|| get_max_tokens(name)),
             keep_alive: Some(KeepAlive::indefinite()),
             supports_tools,
+            supports_thinking,
         }
     }
 
@@ -98,6 +101,7 @@ pub enum ChatMessage {
     Assistant {
         content: String,
         tool_calls: Option<Vec<OllamaToolCall>>,
+        thinking: Option<String>,
     },
     User {
         content: String,
@@ -140,6 +144,7 @@ pub struct ChatRequest {
     pub keep_alive: KeepAlive,
     pub options: Option<ChatOptions>,
     pub tools: Vec<OllamaTool>,
+    pub think: Option<bool>,
 }
 
 impl ChatRequest {
@@ -215,6 +220,10 @@ impl ModelShow {
         // .contains expects &String, which would require an additional allocation
         self.capabilities.iter().any(|v| v == "tools")
     }
+
+    pub fn supports_thinking(&self) -> bool {
+        self.capabilities.iter().any(|v| v == "thinking")
+    }
 }
 
 pub async fn complete(
@@ -459,9 +468,11 @@ mod tests {
             ChatMessage::Assistant {
                 content,
                 tool_calls,
+                thinking,
             } => {
                 assert!(content.is_empty());
                 assert!(tool_calls.is_some_and(|v| !v.is_empty()));
+                assert!(thinking.is_none());
             }
             _ => panic!("Deserialized wrong role"),
         }