feat: add Mallard AI as a model provider

core/llm/autodetect.ts

@@ -61,6 +61,7 @@ const PROVIDER_HANDLES_TEMPLATING: string[] = [
   "sambanova",
   "vertexai",
   "watsonx",
+  "mallard",
   "nebius",
   "relace",
   "openrouter",
@@ -128,6 +129,7 @@ const PROVIDER_SUPPORTS_IMAGES: string[] = [
   "vertexai",
   "azure",
   "scaleway",
+  "mallard",
   "nebius",
   "ovhcloud",
   "watsonx",
@@ -247,6 +249,7 @@ const PARALLEL_PROVIDERS: string[] = [
   "novita",
   "sambanova",
   "ovhcloud",
+  "mallard",
   "nebius",
   "vertexai",
   "function-network",

core/llm/llms/Mallard.ts

@@ -0,0 +1,13 @@
+import { LLMOptions } from "../..";
+
+import OpenAI from "./OpenAI";
+
+class Mallard extends OpenAI {
+  static providerName = "mallard";
+  static defaultOptions: Partial<LLMOptions> = {
+    apiBase: "https://api.mallardai.co.uk/v1/",
+    useLegacyCompletionsEndpoint: false,
+  };
+}
+
+export default Mallard;

core/llm/llms/index.ts

@@ -43,6 +43,7 @@ import MockLLM from "./Mock";
 import Moonshot from "./Moonshot";
 import Msty from "./Msty";
 import NCompass from "./NCompass";
+import Mallard from "./Mallard";
 import Nebius from "./Nebius";
 import Nous from "./Nous";
 import Novita from "./Novita";
@@ -118,6 +119,7 @@ export const LLMClasses = [
   TestLLM,
   Cerebras,
   Asksage,
+  Mallard,
   Nebius,
   Nous,
   Venice,

docs/customize/model-providers/more/mallard.mdx

@@ -0,0 +1,70 @@
+---
+title: "Mallard AI"
+description: "Configure Mallard AI with Continue to use open models through an OpenAI-compatible API, processed only in the UK with zero data retention"
+---
+
+<Info>
+  Mallard AI serves open models through an OpenAI-compatible API, processed
+  only in the UK with zero data retention. Create an account and an API key at
+  [mallardai.co.uk](https://mallardai.co.uk/account/). The [quick
+  start](https://mallardai.co.uk/docs/) covers the API.
+</Info>
+
+## Chat Model
+
+We recommend configuring **Qwen3.6 35B A3B** as your chat model: 262K context window, tool calling and image input.
+
+<Tabs>
+    <Tab title="YAML">
+    ```yaml title="config.yaml"
+    name: My Config
+    version: 0.0.1
+    schema: v1
+
+    models:
+      - name: Qwen3.6 35B A3B
+        provider: mallard
+        model: qwen/qwen3.6-35b-a3b
+        apiKey: <YOUR_MALLARD_API_KEY>
+        capabilities:
+          - tool_use
+          - image_input
+    ```
+    </Tab>
+    <Tab title="JSON">
+    ```json title="config.json"
+    {
+      "models": [
+        {
+          "title": "Qwen3.6 35B A3B",
+          "provider": "mallard",
+          "model": "qwen/qwen3.6-35b-a3b",
+          "apiKey": "<YOUR_MALLARD_API_KEY>"
+        }
+      ]
+    }
+    ```
+    </Tab>
+</Tabs>
+
+## Reasoning
+
+Reasoning is on by default. To switch it off, send `reasoning.enabled: false` with each request:
+
+```yaml title="config.yaml"
+models:
+  - name: Qwen3.6 35B A3B
+    provider: mallard
+    model: qwen/qwen3.6-35b-a3b
+    apiKey: <YOUR_MALLARD_API_KEY>
+    requestOptions:
+      extraBodyProperties:
+        reasoning:
+          enabled: false
+```
+
+## Autocomplete Model
+
+Mallard AI doesn't offer an autocomplete model.
+
+[Click here](../../model-roles/autocomplete) to see a list of autocomplete model providers.

docs/docs.json

@@ -112,6 +112,7 @@
                       "customize/model-providers/more/groq",
                       "customize/model-providers/more/llamacpp",
                       "customize/model-providers/more/llamastack",
+                      "customize/model-providers/more/mallard",
                       "customize/model-providers/more/mimo",
                       "customize/model-providers/more/mistral",
                       "customize/model-providers/more/moonshot",

gui/src/pages/AddNewModel/configs/models.ts

@@ -2052,6 +2052,20 @@ export const models: { [key: string]: ModelPackage } = {
     providerOptions: ["ovhcloud"],
     isOpenSource: true,
   },
+  qwen36_35b_a3b: {
+    title: "Qwen3.6 35B A3B",
+    description:
+      "Qwen3.6-35B-A3B is Qwen's open mixture-of-experts model for agents and coding, with image input and a 262K context window.",
+    refUrl: "https://huggingface.co/Qwen/Qwen3.6-35B-A3B",
+    params: {
+      title: "Qwen3.6 35B A3B",
+      model: "qwen/qwen3.6-35b-a3b",
+      contextLength: 262144,
+    },
+    icon: "qwen.png",
+    providerOptions: ["mallard"],
+    isOpenSource: true,
+  },
   MistralSmall32: {
     title: "Mistral Small 3.2 24B",
     description:

gui/src/pages/AddNewModel/configs/providers.ts

@@ -1182,6 +1182,30 @@ To get started, [register](https://dataplatform.cloud.ibm.com/registration/stepo
     ],
     apiKeyUrl: "https://chat.asksage.ai/",
   },
+  mallard: {
+    title: "Mallard AI",
+    provider: "mallard",
+    refPage: "mallard",
+    description: "Open models processed only in the UK, with zero data retention",
+    longDescription: `Mallard AI serves open models through an OpenAI-compatible API, processed only in the UK with zero data retention. To get started with Mallard AI:\n1. Create an account and an API key at [mallardai.co.uk](https://mallardai.co.uk/account/)\n2. Paste it below\n3. Select a model preset`,
+    params: {
+      apiKey: "",
+    },
+    collectInputFor: [
+      {
+        inputType: "text",
+        key: "apiKey",
+        label: "API Key",
+        placeholder: "Enter your Mallard AI API key",
+        required: true,
+      },
+      ...completionParamsInputsConfigs,
+    ],
+    icon: "mallard.png",
+    tags: [ModelProviderTags.RequiresApiKey, ModelProviderTags.OpenSource],
+    packages: [models.qwen36_35b_a3b],
+    apiKeyUrl: "https://mallardai.co.uk/account/",
+  },
   nebius: {
     title: "Nebius AI Studio",
     provider: "nebius",