diff --git a/docs/src/content/docs/getting-started/configuration.mdx b/docs/src/content/docs/getting-started/configuration.mdx
index b38e054bfa..5d818799bc 100644
--- a/docs/src/content/docs/getting-started/configuration.mdx
+++ b/docs/src/content/docs/getting-started/configuration.mdx
@@ -1078,6 +1078,63 @@ script({
 })
 ```
 
+## LMStudio
+
+The `lmstudio` provider connects to the [LMStudio](https://lmstudio.ai/) headless server.
+and allows to run local LLMs.
+
+<Steps>
+
+<ol>
+
+<li>
+
+Install [LMStudio](https://lmstudio.ai/download) (v0.3.5+)
+
+</li>
+
+<li>
+
+Open LMStudio,
+
+</li>
+
+<li>
+
+Open the [Model Catalog](https://lmstudio.ai/models),
+select your model and load it at least once so it is downloaded locally.
+
+</li>
+
+<li>
+
+Open the settings (Gearwheel icon) and enable **Enable Local LLM Service**.
+
+</li>
+
+<li>
+
+GenAIScript assumes the local server is at `http://localhost:1234/v1` by default.
+Add a `LMSTUDIO_API_BASE` environment variable to change the server URL.
+
+```txt title=".env"
+LMSTUDIO_API_BASE=http://localhost:2345/v1
+```
+
+</li>
+
+</ol>
+
+</Steps>
+
+Find the model **API identifier** in the dialog of loaded models then use that identifier in your script:
+
+```js '"lmstudio:llama-3.2-1b"'
+script({
+    model: "lmstudio:llama-3.2-1b-instruct",
+})
+```
+
 ## LocalAI
 
 [LocalAI](https://localai.io/) act as a drop-in replacement REST API that’s compatible
@@ -1118,9 +1175,9 @@ that allows you to run an LLM locally.
 
 The provider is `llamafile` and the model name is ignored.
 
-## Jan, LMStudio, LLaMA.cpp
+## Jan, LLaMA.cpp
 
-[Jan](https://jan.ai/), [LMStudio](https://lmstudio.ai/),
+[Jan](https://jan.ai/),
 [LLaMA.cpp](https://github.com/ggerganov/llama.cpp/tree/master/examples/server)
 also allow running models locally or interfacing with other LLM vendors.
 
diff --git a/packages/core/src/connection.ts b/packages/core/src/connection.ts
index bc43bef4d7..1cd7873d61 100644
--- a/packages/core/src/connection.ts
+++ b/packages/core/src/connection.ts
@@ -30,6 +30,8 @@ import {
     ALIBABA_BASE,
     MODEL_PROVIDER_MISTRAL,
     MISTRAL_API_BASE,
+    MODEL_PROVIDER_LMSTUDIO,
+    LMSTUDIO_API_BASE,
 } from "./constants"
 import { fileExists, readText, writeText } from "./fs"
 import {
@@ -427,10 +429,14 @@ export async function parseTokenFromEnv(
     }
 
     if (provider === MODEL_PROVIDER_LLAMAFILE) {
+        const base =
+            findEnvVar(env, "LLAMAFILE", BASE_SUFFIX)?.value ||
+            LLAMAFILE_API_BASE
+        if (!URL.canParse(base)) throw new Error(`${base} must be a valid URL`)
         return {
             provider,
             model,
-            base: LLAMAFILE_API_BASE,
+            base,
             token: "llamafile",
             type: "openai",
             source: "default",
@@ -438,16 +444,33 @@ export async function parseTokenFromEnv(
     }
 
     if (provider === MODEL_PROVIDER_LITELLM) {
+        const base =
+            findEnvVar(env, "LITELLM", BASE_SUFFIX)?.value || LITELLM_API_BASE
+        if (!URL.canParse(base)) throw new Error(`${base} must be a valid URL`)
         return {
             provider,
             model,
-            base: LITELLM_API_BASE,
+            base,
             token: "litellm",
             type: "openai",
             source: "default",
         }
     }
 
+    if (provider === MODEL_PROVIDER_LMSTUDIO) {
+        const base =
+            findEnvVar(env, "LMSTUDIO", BASE_SUFFIX)?.value || LMSTUDIO_API_BASE
+        if (!URL.canParse(base)) throw new Error(`${base} must be a valid URL`)
+        return {
+            provider,
+            model,
+            base,
+            token: "lmstudio",
+            type: "openai",
+            source: "env: LMSTUDIO_API_...",
+        }
+    }
+
     if (provider === MODEL_PROVIDER_TRANSFORMERS) {
         return {
             provider,
diff --git a/packages/core/src/constants.ts b/packages/core/src/constants.ts
index 3a9c4ef6da..0c367aeb53 100644
--- a/packages/core/src/constants.ts
+++ b/packages/core/src/constants.ts
@@ -155,6 +155,7 @@ export const GOOGLE_API_BASE =
 export const ALIBABA_BASE =
     "https://dashscope-intl.aliyuncs.com/compatible-mode/v1"
 export const MISTRAL_API_BASE = "https://api.mistral.ai/v1"
+export const LMSTUDIO_API_BASE = "http://localhost:1234/v1"
 
 export const PROMPTFOO_CACHE_PATH = ".genaiscript/cache/tests"
 export const PROMPTFOO_CONFIG_DIR = ".genaiscript/config/tests"
@@ -184,6 +185,7 @@ export const MODEL_PROVIDER_HUGGINGFACE = "huggingface"
 export const MODEL_PROVIDER_TRANSFORMERS = "transformers"
 export const MODEL_PROVIDER_ALIBABA = "alibaba"
 export const MODEL_PROVIDER_MISTRAL = "mistral"
+export const MODEL_PROVIDER_LMSTUDIO = "lmstudio"
 
 export const TRACE_FILE_PREVIEW_MAX_LENGTH = 240
 
@@ -208,6 +210,8 @@ export const DOCS_CONFIGURATION_AZURE_MODELS_SERVERLESS_URL =
     "https://microsoft.github.io/genaiscript/getting-started/configuration/#azure_serverless_models"
 export const DOCS_CONFIGURATION_OLLAMA_URL =
     "https://microsoft.github.io/genaiscript/getting-started/configuration/#ollama"
+export const DOCS_CONFIGURATION_LMSTUDIO_URL =
+    "https://microsoft.github.io/genaiscript/getting-started/configuration/#lmstudio"
 export const DOCS_CONFIGURATION_LLAMAFILE_URL =
     "https://microsoft.github.io/genaiscript/getting-started/configuration/#llamafile"
 export const DOCS_CONFIGURATION_LITELLM_URL =
@@ -295,6 +299,11 @@ export const MODEL_PROVIDERS = Object.freeze([
         detail: "Ollama local model",
         url: DOCS_CONFIGURATION_OLLAMA_URL,
     },
+    {
+        id: MODEL_PROVIDER_LMSTUDIO,
+        detail: "LM Studio local server",
+        url: DOCS_CONFIGURATION_LMSTUDIO_URL,
+    },
     {
         id: MODEL_PROVIDER_ALIBABA,
         detail: "Alibaba models",
diff --git a/packages/vscode/src/lmaccess.ts b/packages/vscode/src/lmaccess.ts
index 46dcd5ed3c..bea3b4af50 100644
--- a/packages/vscode/src/lmaccess.ts
+++ b/packages/vscode/src/lmaccess.ts
@@ -16,6 +16,7 @@ import {
     DOCS_CONFIGURATION_URL,
     MODEL_PROVIDER_GOOGLE,
     MODEL_PROVIDER_ALIBABA,
+    MODEL_PROVIDER_LMSTUDIO,
 } from "../../core/src/constants"
 import { OpenAIAPIType } from "../../core/src/host"
 import { parseModelIdentifier } from "../../core/src/models"
@@ -39,6 +40,7 @@ async function generateLanguageModelConfiguration(
         MODEL_PROVIDER_AZURE_SERVERLESS_OPENAI,
         MODEL_PROVIDER_AZURE_SERVERLESS_MODELS,
         MODEL_PROVIDER_LITELLM,
+        MODEL_PROVIDER_LMSTUDIO,
         MODEL_PROVIDER_GOOGLE,
         MODEL_PROVIDER_ALIBABA,
     ]