diff --git a/README.md b/README.md
index 574fdc8..4e7949f 100644
--- a/README.md
+++ b/README.md
@@ -78,6 +78,10 @@ model = "google/gemini-3-flash"
 model = "gemma-4-E2B-it.litertlm"
 huggingface_repo = "litert-community/gemma-4-E2B-it-litert-lm"
 
+[lmstudio]
+model = "qwen/qwen3-coder-next"
+# api_url = "http://localhost:1234"  # default, change if running elsewhere
+
 [apple_intelligence]
 model = "system"
 ```
diff --git a/src/config.rs b/src/config.rs
index 771fd70..93b1178 100644
--- a/src/config.rs
+++ b/src/config.rs
@@ -10,6 +10,7 @@ pub struct Config {
     pub anthropic: Option<ProviderConfig>,
     pub openai: Option<ProviderConfig>,
     pub google: Option<ProviderConfig>,
+    pub lmstudio: Option<ProviderConfig>,
     pub mistral: Option<ProviderConfig>,
     pub ollama: Option<ProviderConfig>,
     pub openrouter: Option<ProviderConfig>,
@@ -101,6 +102,10 @@ model = "google/gemini-3-flash"
 model = "gemma-4-E2B-it.litertlm"
 huggingface_repo = "litert-community/gemma-4-E2B-it-litert-lm"
 
+[lmstudio]
+model = "qwen/qwen3-coder-next"
+# api_url = "http://localhost:1234"  # default, change if running elsewhere
+
 # On-device Apple Intelligence (FoundationModels framework).
 # Only used when commandok was built with: cargo install commandok --features apple-intelligence
 # Requires macOS 26+ on Apple Silicon and Apple Intelligence enabled in System Settings.
@@ -165,6 +170,7 @@ const PROVIDER_ORDER: &[&str] = &[
     "anthropic",
     "openai",
     "google",
+    "lmstudio",
     "mistral",
     "ollama",
     "openrouter",
@@ -185,6 +191,7 @@ impl Config {
             "anthropic" => self.anthropic.as_ref(),
             "openai" => self.openai.as_ref(),
             "google" => self.google.as_ref(),
+            "lmstudio" => self.lmstudio.as_ref(),
             "mistral" => self.mistral.as_ref(),
             "ollama" => self.ollama.as_ref(),
             "openrouter" => self.openrouter.as_ref(),
diff --git a/src/provider/lmstudio.rs b/src/provider/lmstudio.rs
new file mode 100644
index 0000000..c5a1180
--- /dev/null
+++ b/src/provider/lmstudio.rs
@@ -0,0 +1,64 @@
+use super::ApiEvent;
+use crate::config::ProviderConfig;
+use tokio::sync::mpsc;
+
+pub async fn stream(
+    cfg: &ProviderConfig,
+    query: &str,
+    system_prompt: &str,
+    tx: mpsc::UnboundedSender<ApiEvent>,
+) {
+    let client = reqwest::Client::new();
+    let base_url = if cfg.api_url.is_empty() {
+        "http://localhost:1234"
+    } else {
+        cfg.api_url.trim_end_matches('/')
+    };
+    let url = format!("{base_url}/api/v1/chat");
+
+    let body = serde_json::json!({
+        "model": cfg.model,
+        "system_prompt": system_prompt,
+        "input": query,
+    });
+
+    let request = client
+        .post(&url)
+        .header("content-type", "application/json")
+        .json(&body);
+
+    let Some(resp) = super::send_request(request, &tx).await else {
+        return;
+    };
+
+    let json = match resp.json::<serde_json::Value>().await {
+        Ok(j) => j,
+        Err(e) => {
+            let _ = tx.send(ApiEvent::Error(e.to_string()));
+            return;
+        }
+    };
+
+    let outputs = json["output"].as_array();
+    let has_content = outputs
+        .map(|arr| arr.iter().any(|item| item["content"].as_str().is_some()))
+        .unwrap_or(false);
+
+    if !has_content {
+        let err_msg = json["error"]
+            .as_str()
+            .unwrap_or("invalid response: output[].content missing");
+        let _ = tx.send(ApiEvent::Error(err_msg.to_string()));
+        return;
+    }
+
+    for item in outputs.unwrap() {
+        if let Some(text) = item["content"].as_str() {
+            if !text.is_empty() && tx.send(ApiEvent::Delta(text.to_string())).is_err() {
+                return;
+            }
+        }
+    }
+
+    let _ = tx.send(ApiEvent::Done);
+}
diff --git a/src/provider/mod.rs b/src/provider/mod.rs
index 45e31f6..08afc0c 100644
--- a/src/provider/mod.rs
+++ b/src/provider/mod.rs
@@ -7,6 +7,7 @@ pub mod apple_intelligence;
 pub mod claude;
 pub mod gemini;
 pub mod litert_lm;
+pub mod lmstudio;
 pub mod mistral;
 pub mod ollama;
 pub mod openai;
@@ -28,6 +29,7 @@ pub enum Provider {
     Anthropic(ProviderConfig),
     OpenAi(ProviderConfig),
     Google(ProviderConfig),
+    LMStudio(ProviderConfig),
     Mistral(ProviderConfig),
     Ollama(ProviderConfig),
     OpenRouter(ProviderConfig),
@@ -48,6 +50,7 @@ impl Provider {
             "anthropic" => Provider::Anthropic(cfg.clone()),
             "openai" => Provider::OpenAi(cfg.clone()),
             "google" => Provider::Google(cfg.clone()),
+            "lmstudio" => Provider::LMStudio(cfg.clone()),
             "mistral" => Provider::Mistral(cfg.clone()),
             "ollama" => Provider::Ollama(cfg.clone()),
             "openrouter" => Provider::OpenRouter(cfg.clone()),
@@ -74,6 +77,7 @@ impl Provider {
             Provider::Anthropic(cfg) => claude::stream(cfg, query, system_prompt, tx).await,
             Provider::OpenAi(cfg) => openai::stream(cfg, query, system_prompt, tx).await,
             Provider::Google(cfg) => gemini::stream(cfg, query, system_prompt, tx).await,
+            Provider::LMStudio(cfg) => lmstudio::stream(cfg, query, system_prompt, tx).await,
             Provider::Mistral(cfg) => mistral::stream(cfg, query, system_prompt, tx).await,
             Provider::Ollama(cfg) => ollama::stream(cfg, query, system_prompt, tx).await,
             Provider::OpenRouter(cfg) => openrouter::stream(cfg, query, system_prompt, tx).await,