feat: add Mixlayer as an LLM provider

core/llm/autodetect.ts

@@ -68,6 +68,7 @@ const PROVIDER_HANDLES_TEMPLATING: string[] = [
   "deepseek",
   "xAI",
   "minimax",
+  "mixlayer",
   "groq",
   "gemini",
   "docker",
@@ -133,6 +134,7 @@ const PROVIDER_SUPPORTS_IMAGES: string[] = [
   "watsonx",
   "zAI",
   "tensorix",
+  "mixlayer",
 ];
 
 const MODEL_SUPPORTS_IMAGES: RegExp[] = [
@@ -153,6 +155,10 @@ const MODEL_SUPPORTS_IMAGES: RegExp[] = [
   /\bgemma-?[34](?!n)/, // gemma3/gemma4 support vision, but gemma3n doesn't!
   /\b(pali|med)gemma/,
   /qwen(.*)vl/,
+  /qwen\/qwen3\.8-27b$/,
+  /qwen\/qwen3\.6-35b-a3b$/,
+  /qwen\/qwen3\.5-9b$/,
+  /qwen\/qwen3\.5-4b-free$/,
   /mistral-small/,
   /mistral-medium/,
 ];
@@ -252,6 +258,7 @@ const PARALLEL_PROVIDERS: string[] = [
   "function-network",
   "scaleway",
   "minimax",
+  "mixlayer",
   "tensorix",
 ];
 

core/llm/llms/Mixlayer.ts

@@ -0,0 +1,14 @@
+import { LLMOptions } from "../../index.js";
+
+import OpenAI from "./OpenAI.js";
+
+class Mixlayer extends OpenAI {
+  static providerName = "mixlayer";
+  static defaultOptions: Partial<LLMOptions> = {
+    apiBase: "https://mixlayer.ai/v1/",
+    model: "z-ai/glm-5.3",
+    useLegacyCompletionsEndpoint: false,
+  };
+}
+
+export default Mixlayer;

core/llm/llms/OpenAI-compatible.vitest.ts

@@ -10,6 +10,7 @@ import OpenRouter from "./OpenRouter.js";
 import xAI from "./xAI.js";
 import Mistral from "./Mistral.js";
 import Mimo from "./Mimo.js";
+import Mixlayer from "./Mixlayer.js";
 import LMStudio from "./LMStudio.js";
 import Cerebras from "./Cerebras.js";
 import DeepInfra from "./DeepInfra.js";
@@ -306,6 +307,11 @@ createOpenAISubclassTests(Mimo, {
   defaultApiBase: "https://api.xiaomimimo.com/v1/",
 });
 
+createOpenAISubclassTests(Mixlayer, {
+  providerName: "mixlayer",
+  defaultApiBase: "https://mixlayer.ai/v1/",
+});
+
 createOpenAISubclassTests(LMStudio, {
   providerName: "lmstudio",
   defaultApiBase: "http://localhost:1234/v1/",

core/llm/llms/index.ts

@@ -39,6 +39,7 @@ import LMStudio from "./LMStudio";
 import Mistral from "./Mistral";
 import Mimo from "./Mimo";
 import MiniMax from "./MiniMax";
+import Mixlayer from "./Mixlayer";
 import MockLLM from "./Mock";
 import Moonshot from "./Moonshot";
 import Msty from "./Msty";
@@ -95,6 +96,7 @@ export const LLMClasses = [
   Mistral,
   Mimo,
   MiniMax,
+  Mixlayer,
   Bedrock,
   BedrockImport,
   SageMaker,

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

@@ -0,0 +1,62 @@
+---
+title: "How to Configure Mixlayer with Continue"
+sidebarTitle: "Mixlayer"
+---
+
+<Info>
+  Get your API key from the [Mixlayer console](https://console.mixlayer.com/app/api-keys)
+</Info>
+
+[Mixlayer](https://mixlayer.com) serves open-source models (Qwen, DeepSeek, GLM) behind an OpenAI-compatible API, with zero data retention.
+
+## Configuration
+
+<Tabs>
+  <Tab title="YAML">
+  ```yaml title="config.yaml"
+  name: My Config
+  version: 0.0.1
+  schema: v1
+
+  models:
+    - name: GLM 5.3
+      provider: mixlayer
+      model: z-ai/glm-5.3
+      apiKey: ${{ secrets.MIXLAYER_API_KEY }}
+  ```
+  </Tab>
+  <Tab title="JSON (Deprecated)">
+  ```json title="config.json"
+  {
+    "models": [
+      {
+        "title": "GLM 5.3",
+        "provider": "mixlayer",
+        "model": "z-ai/glm-5.3",
+        "apiKey": "<YOUR_MIXLAYER_API_KEY>"
+      }
+    ]
+  }
+  ```
+  </Tab>
+</Tabs>
+
+## Available Models
+
+| Model | Context | Input | Notes |
+| :---- | :------ | :---- | :---- |
+| `z-ai/glm-5.3` | 262,144 | text | Tools. Reasoning is always on (Low/High/Max; `none` is rejected). |
+| `qwen/qwen3.8-27b` | 262,144 | text, image | Tools. Optional thinking (default on; `none` disables). |
+| `qwen/qwen3.6-35b-a3b` | 131,072 | text, image | Tools. Optional thinking (default on; `none` disables). |
+| `deepseek/deepseek-v4.1-flash` | 262,144 | text | Tools. Optional thinking (default High; `none` disables). |
+| `qwen/qwen3.5-9b` | 131,072 | text, image | Tools. Optional thinking (default on; `none` disables). |
+| `qwen/qwen3.5-4b-free` | 131,072 | text, image | Free. Tools. Optional thinking (default on; `none` disables). |
+
+## Notes
+
+- OpenAI-compatible API at `https://mixlayer.ai/v1` (nearest edge). Set `apiBase` to `https://<region>.mixlayer.ai/v1` for a specific region.
+- In `config.yaml`, reference the key with `${{ secrets.MIXLAYER_API_KEY }}`. Continue resolves that from a `.env` file, or from the process environment when using the CLI.
+- Prompts and outputs are processed in memory and discarded after inference (zero data retention). They are not stored or used for training.
+- Chat Completions does not support `tool_choice`.
+- Max output tokens per model are not published.
+- Pricing: [mixlayer.com/pricing](https://mixlayer.com/pricing)

docs/customize/model-providers/overview.mdx

@@ -40,6 +40,7 @@ Beyond the top-level providers, Continue supports many other options:
 | [NVIDIA](/customize/model-providers/more/nvidia)                       | GPU-accelerated model hosting                              |
 | [Cloudflare](/customize/model-providers/more/cloudflare)               | Edge-based AI inference services                           |
 | [MiniMax](/customize/model-providers/more/minimax)                     | High-performance models with 200K+ context window          |
+| [Mixlayer](/customize/model-providers/more/mixlayer)                   | OpenAI-compatible inference for open-source models, with zero data retention |
 
 ### Local Model Options
 

docs/docs.json

@@ -114,6 +114,7 @@
                       "customize/model-providers/more/llamastack",
                       "customize/model-providers/more/mimo",
                       "customize/model-providers/more/mistral",
+                      "customize/model-providers/more/mixlayer",
                       "customize/model-providers/more/moonshot",
                       "customize/model-providers/more/nous",
                       "customize/model-providers/more/nvidia",

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

@@ -3327,4 +3327,78 @@ export const models: { [key: string]: ModelPackage } = {
     icon: "inception.png",
     isOpenSource: false,
   },
+
+  // Mixlayer models
+  mixlayerGlm53: {
+    title: "GLM 5.3",
+    description:
+      "Text model with tool calling. Reasoning is always on. 262,144 token context window.",
+    params: {
+      title: "GLM 5.3",
+      model: "z-ai/glm-5.3",
+      contextLength: 262_144,
+    },
+    providerOptions: ["mixlayer"],
+    isOpenSource: true,
+  },
+  mixlayerQwen3827b: {
+    title: "Qwen3.8 27B",
+    description:
+      "Text and image model with tool calling and optional thinking. 262,144 token context window.",
+    params: {
+      title: "Qwen3.8 27B",
+      model: "qwen/qwen3.8-27b",
+      contextLength: 262_144,
+    },
+    providerOptions: ["mixlayer"],
+    isOpenSource: true,
+  },
+  mixlayerQwen3635bA3b: {
+    title: "Qwen3.6 35B A3B",
+    description:
+      "Text and image model with tool calling and optional thinking. 131,072 token context window.",
+    params: {
+      title: "Qwen3.6 35B A3B",
+      model: "qwen/qwen3.6-35b-a3b",
+      contextLength: 131_072,
+    },
+    providerOptions: ["mixlayer"],
+    isOpenSource: true,
+  },
+  mixlayerDeepseekV41Flash: {
+    title: "DeepSeek V4.1 Flash",
+    description:
+      "Text model with tool calling and optional thinking. 262,144 token context window.",
+    params: {
+      title: "DeepSeek V4.1 Flash",
+      model: "deepseek/deepseek-v4.1-flash",
+      contextLength: 262_144,
+    },
+    providerOptions: ["mixlayer"],
+    isOpenSource: true,
+  },
+  mixlayerQwen359b: {
+    title: "Qwen3.5 9B",
+    description:
+      "Text and image model with tool calling and optional thinking. 131,072 token context window.",
+    params: {
+      title: "Qwen3.5 9B",
+      model: "qwen/qwen3.5-9b",
+      contextLength: 131_072,
+    },
+    providerOptions: ["mixlayer"],
+    isOpenSource: true,
+  },
+  mixlayerQwen354bFree: {
+    title: "Qwen3.5 4B (Free)",
+    description:
+      "Free text and image model with tool calling and optional thinking. 131,072 token context window.",
+    params: {
+      title: "Qwen3.5 4B (Free)",
+      model: "qwen/qwen3.5-4b-free",
+      contextLength: 131_072,
+    },
+    providerOptions: ["mixlayer"],
+    isOpenSource: true,
+  },
 };

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

@@ -589,6 +589,40 @@ Select the \`GPT-4o\` model below to complete your provider configuration, but n
     ],
     apiKeyUrl: "https://platform.minimax.io",
   },
+  mixlayer: {
+    title: "Mixlayer",
+    provider: "mixlayer",
+    description:
+      "OpenAI-compatible inference for open-source models, with zero data retention.",
+    longDescription:
+      "Mixlayer serves Qwen, DeepSeek, and GLM models at https://mixlayer.ai/v1. Get an API key from the [Mixlayer console](https://console.mixlayer.com/app/api-keys). A free Qwen 3.5 4B model is available.",
+    tags: [ModelProviderTags.RequiresApiKey],
+    collectInputFor: [
+      {
+        inputType: "text",
+        key: "apiKey",
+        label: "API Key",
+        placeholder: "Enter your Mixlayer API key",
+        required: true,
+      },
+    ],
+    packages: [
+      models.mixlayerGlm53,
+      models.mixlayerQwen3827b,
+      models.mixlayerQwen3635bA3b,
+      models.mixlayerDeepseekV41Flash,
+      models.mixlayerQwen359b,
+      models.mixlayerQwen354bFree,
+      {
+        ...models.AUTODETECT,
+        params: {
+          ...models.AUTODETECT.params,
+          title: "Mixlayer",
+        },
+      },
+    ],
+    apiKeyUrl: "https://console.mixlayer.com/app/api-keys",
+  },
   inception: {
     title: "Inception Labs",
     provider: "inception",

packages/config-types/src/index.ts

@@ -61,6 +61,7 @@ export const modelDescriptionSchema = z.object({
     "scaleway",
     "watsonx",
     "minimax",
+    "mixlayer",
   ]),
   model: z.string(),
   apiKey: z.string().optional(),

packages/llm-info/src/index.ts

@@ -6,6 +6,7 @@ import { CometAPI } from "./providers/cometapi.js";
 import { Gemini } from "./providers/gemini.js";
 import { Inception } from "./providers/inception.js";
 import { MiniMax } from "./providers/minimax.js";
+import { Mixlayer } from "./providers/mixlayer.js";
 import { Mistral } from "./providers/mistral.js";
 import { Ollama } from "./providers/ollama.js";
 import { OpenAi } from "./providers/openai.js";
@@ -29,6 +30,7 @@ export const allModelProviders: ModelProvider[] = [
   CometAPI,
   Inception,
   MiniMax,
+  Mixlayer,
   xAI,
   zAI,
 ];

packages/llm-info/src/providers/mixlayer.ts

@@ -0,0 +1,66 @@
+import { MediaType, ModelProvider } from "../types.js";
+
+export const Mixlayer: ModelProvider = {
+  models: [
+    {
+      model: "z-ai/glm-5.3",
+      displayName: "GLM 5.3",
+      contextLength: 262144,
+      description:
+        "Text model with tool calling. Reasoning is always on (Low/High/Max; none is rejected).",
+      regex: /^z-ai\/glm-5\.3$/,
+      recommendedFor: ["chat"],
+    },
+    {
+      model: "qwen/qwen3.8-27b",
+      displayName: "Qwen3.8 27B",
+      contextLength: 262144,
+      description:
+        "Text and image model with tool calling. Optional thinking (default on; none disables).",
+      regex: /^qwen\/qwen3\.8-27b$/,
+      mediaTypes: [MediaType.Text, MediaType.Image],
+      recommendedFor: ["chat"],
+    },
+    {
+      model: "qwen/qwen3.6-35b-a3b",
+      displayName: "Qwen3.6 35B A3B",
+      contextLength: 131072,
+      description:
+        "Text and image model with tool calling. Optional thinking (default on; none disables).",
+      regex: /^qwen\/qwen3\.6-35b-a3b$/,
+      mediaTypes: [MediaType.Text, MediaType.Image],
+      recommendedFor: ["chat"],
+    },
+    {
+      model: "deepseek/deepseek-v4.1-flash",
+      displayName: "DeepSeek V4.1 Flash",
+      contextLength: 262144,
+      description:
+        "Text model with tool calling. Optional thinking (default High; none disables).",
+      regex: /^deepseek\/deepseek-v4\.1-flash$/,
+      recommendedFor: ["chat"],
+    },
+    {
+      model: "qwen/qwen3.5-9b",
+      displayName: "Qwen3.5 9B",
+      contextLength: 131072,
+      description:
+        "Text and image model with tool calling. Optional thinking (default on; none disables).",
+      regex: /^qwen\/qwen3\.5-9b$/,
+      mediaTypes: [MediaType.Text, MediaType.Image],
+      recommendedFor: ["chat"],
+    },
+    {
+      model: "qwen/qwen3.5-4b-free",
+      displayName: "Qwen3.5 4B (Free)",
+      contextLength: 131072,
+      description:
+        "Free text and image model with tool calling. Optional thinking (default on; none disables).",
+      regex: /^qwen\/qwen3\.5-4b-free$/,
+      mediaTypes: [MediaType.Text, MediaType.Image],
+      recommendedFor: ["chat"],
+    },
+  ],
+  id: "mixlayer",
+  displayName: "Mixlayer",
+};

packages/openai-adapters/src/index.ts

@@ -135,6 +135,8 @@ export function constructLlmApi(config: LLMConfig): BaseLlmApi | undefined {
       return openAICompatible("https://api.voyageai.com/v1/", config);
     case "mistral":
       return openAICompatible("https://api.mistral.ai/v1/", config);
+    case "mixlayer":
+      return openAICompatible("https://mixlayer.ai/v1/", config);
     case "deepinfra":
       return openAICompatible("https://api.deepinfra.com/v1/openai/", config);
     case "vllm":

packages/openai-adapters/src/types.ts

@@ -34,6 +34,7 @@ export const OpenAIConfigSchema = BasePlusConfig.extend({
   provider: z.union([
     z.literal("openai"),
     z.literal("mistral"),
+    z.literal("mixlayer"),
     z.literal("voyage"),
     z.literal("deepinfra"),
     z.literal("groq"),