feat(llm): add NEAR AI Cloud provider

.env.example

@@ -41,6 +41,15 @@ CLAWDBOT_AUTH_TOKEN=your-openclaw-gateway-token
 # Session key prefix — change if running multiple instances on the same gateway
 GATEWAY_SESSION_KEY=voice-main-1
 
+# =============================================================================
+# LLM provider keys (optional — configured through OpenClaw / admin model config)
+# =============================================================================
+
+# NEAR AI Cloud (OpenAI-compatible TEE inference)
+# Get key: https://cloud.near.ai
+# NEARAI_API_KEY=your-nearai-api-key
+# NEARAI_BASE_URL=https://cloud-api.near.ai/v1
+
 # =============================================================================
 # TTS — choose one or more providers
 # =============================================================================

README.md

@@ -157,7 +157,7 @@ See [`deploy/`](deploy/) for the full production setup including SSL, nginx reve
 All configuration is in `.env`. Copy `.env.example` to `.env` and fill in your values.
 
 **Required:**
-- An LLM provider API key (OpenAI, Anthropic, Groq, Z.AI, or any OpenClaw-compatible provider)
+- An LLM provider API key (OpenAI, Anthropic, Groq, Z.AI, NEAR AI Cloud, or any OpenClaw-compatible provider)
 - `CLAWDBOT_AUTH_TOKEN` — set during `npx openvoiceui setup` or in OpenClaw's setup wizard
 
 **Optional but recommended:**
@@ -177,6 +177,7 @@ See [`.env.example`](.env.example) for all available options with descriptions.
 |----------|--------|
 | OpenClaw Gateway | Built-in — routes to OpenAI, Anthropic, Groq, Z.AI, and more |
 | Z.AI (GLM-5-turbo) | Built-in |
+| NEAR AI Cloud | OpenAI-compatible TEE inference via OpenClaw |
 | Groq (Llama, Qwen) | Via OpenClaw |
 | Google Gemini | Via OpenClaw |
 | MiniMax | Via OpenClaw |

SETUP.md

@@ -93,6 +93,8 @@ GROQ_API_KEY=your-groq-key
 SECRET_KEY=any-random-string-here
 ```
 
+Optional: set `NEARAI_API_KEY=your-nearai-key` to enable NEAR AI Cloud TEE inference.
+
 **Optional: enable the coding-agent skill**
 
 The coding-agent skill lets the AI write code, create files, and run commands
@@ -160,6 +162,8 @@ Point openvoiceui at your running OpenClaw gateway instead of starting a new one
    SECRET_KEY=any-random-string-here
    ```
 
+   Optional: set `NEARAI_API_KEY=your-nearai-key` to enable NEAR AI Cloud TEE inference.
+
 4. Start only the openvoiceui and supertonic services (skip the built-in openclaw):
    ```bash
    docker compose up --build openvoiceui supertonic

config/providers.yaml

@@ -11,6 +11,7 @@ llm:
   fallback_provider: clawdbot
   modules:
     - providers.llm.zai_provider
+    - providers.llm.nearai_provider
     - providers.llm.clawdbot_provider
 
   providers:
@@ -25,6 +26,22 @@ llm:
         - id: glm-4-plus
           name: "GLM-4 Plus"
 
+    nearai:
+      name: "NEAR AI Cloud"
+      api_key: ${NEARAI_API_KEY}
+      base_url: https://cloud-api.near.ai/v1
+      default_model: zai-org/GLM-5.1-FP8
+      priority: 15
+      models:
+        - id: zai-org/GLM-5.1-FP8
+          name: "GLM 5.1"
+        - id: Qwen/Qwen3.6-35B-A3B-FP8
+          name: "Qwen 3.6 35B A3B FP8"
+        - id: Qwen/Qwen3.5-122B-A10B
+          name: "Qwen3.5 122B A10B"
+        - id: Qwen/Qwen3-VL-30B-A3B-Instruct
+          name: "Qwen3-VL 30B A3B Instruct"
+
     clawdbot:
       name: "Clawdbot Gateway"
       gateway_url: ${CLAWDBOT_GATEWAY_URL}

docs/getting-started/configuration.md

@@ -25,6 +25,7 @@ OpenVoiceUI needs at least one LLM provider to power conversations. The AI gatew
 |----------|---------|-------------------|
 | Groq (Llama, Qwen) | `GROQ_API_KEY` | [console.groq.com](https://console.groq.com) |
 | Z.AI (GLM-5-turbo) | Set via OpenClaw config | [z.ai](https://z.ai) |
+| NEAR AI Cloud (TEE inference) | `NEARAI_API_KEY` | [cloud.near.ai](https://cloud.near.ai) |
 | OpenAI (GPT-4o) | `OPENAI_API_KEY` | [platform.openai.com](https://platform.openai.com) |
 | Anthropic (Claude) | `ANTHROPIC_API_KEY` | [console.anthropic.com](https://console.anthropic.com) |
 
@@ -106,6 +107,8 @@ Complete list of all recognized environment variables:
 | `GATEWAY_SESSION_KEY` | No | `voice-main-1` | Session key prefix. Change if running multiple instances on the same gateway. |
 | `OPENCLAW_VERSION` | No | `2026.3.13` | Docker build arg: OpenClaw version to install. Only change if you have verified compatibility. |
 | `CANVAS_PAGES_DIR` | No | `runtime/canvas-pages/` | Where canvas HTML pages are stored. Docker: leave unset (uses volume mount). VPS: set to the path created by the deploy script. |
+| `NEARAI_API_KEY` | No | -- | NEAR AI Cloud API key for OpenAI-compatible TEE inference. |
+| `NEARAI_BASE_URL` | No | `https://cloud-api.near.ai/v1` | Override for the NEAR AI Cloud OpenAI-compatible endpoint. |
 | `GROQ_API_KEY` | Recommended | -- | Groq API key for Orpheus TTS and Whisper STT. |
 | `USE_GROQ` | No | -- | Set `true` to enable Groq integration. |
 | `USE_GROQ_TTS` | No | -- | Set `true` to enable Groq Orpheus TTS. |

docs/getting-started/installation.md

@@ -85,6 +85,8 @@ GROQ_API_KEY=your-groq-api-key
 SECRET_KEY=any-random-string-here
 ```
 
+Optional: set `NEARAI_API_KEY=your-nearai-key` to enable NEAR AI Cloud TEE inference.
+
 See [Configuration](/getting-started/configuration) for the full list of environment variables.
 
 ### 2. Optional: enable coding agent

docs/reference/architecture.md

@@ -446,6 +446,8 @@ Set in `app.py` after_request handler.
 | Variable | Default | Purpose |
 |----------|---------|---------|
 | `CLAWDBOT_GATEWAY_URL` | `ws://127.0.0.1:18791` | Gateway WebSocket URL |
+| `NEARAI_API_KEY` | — | NEAR AI Cloud TEE inference |
+| `NEARAI_BASE_URL` | `https://cloud-api.near.ai/v1` | Optional NEAR AI endpoint override |
 | `GROQ_API_KEY` | — | Groq Orpheus TTS |
 | `FAL_KEY` | — | Qwen3-TTS via fal.ai |
 | `SUNO_API_KEY` | — | AI song generation |

install.js

@@ -71,6 +71,13 @@ module.exports = {
             placeholder: "",
             required: false,
           },
+          {
+            key: "NEARAI_API_KEY",
+            title: "[RECOMMENDED] NEAR AI Cloud API Key — TEE inference",
+            description: "cloud.near.ai — OpenAI-compatible private inference",
+            placeholder: "",
+            required: false,
+          },
           {
             key: "ANTHROPIC_API_KEY",
             title: "[RECOMMENDED] Anthropic API Key — Claude models",
@@ -314,6 +321,7 @@ module.exports = {
           PINOKIO_DEEPGRAM_API_KEY: "{{input.DEEPGRAM_API_KEY}}",
           PINOKIO_ANTHROPIC_API_KEY: "{{input.ANTHROPIC_API_KEY}}",
           PINOKIO_ZAI_API_KEY: "{{input.ZAI_API_KEY}}",
+          PINOKIO_NEARAI_API_KEY: "{{input.NEARAI_API_KEY}}",
           PINOKIO_OPENAI_API_KEY: "{{input.OPENAI_API_KEY}}",
           PINOKIO_GEMINI_API_KEY: "{{input.GEMINI_API_KEY}}",
           PINOKIO_OPENROUTER_API_KEY: "{{input.OPENROUTER_API_KEY}}",

profiles/schema.json

@@ -43,7 +43,7 @@
       "properties": {
         "provider": {
           "type": "string",
-          "enum": ["zai", "clawdbot", "gateway", "openai", "ollama", "anthropic", "hume", "elevenlabs"]
+          "enum": ["zai", "nearai", "clawdbot", "gateway", "openai", "ollama", "anthropic", "hume", "elevenlabs"]
         },
         "model":     { "type": "string" },
         "config_id": { "type": "string" },

providers/llm/__init__.py

@@ -8,5 +8,6 @@
 # Import concrete providers so their registry.register() calls fire
 from providers.llm import zai_provider  # noqa: F401
 from providers.llm import clawdbot_provider  # noqa: F401
+from providers.llm import nearai_provider  # noqa: F401
 
 __all__ = ["LLMProvider", "LLMResponse", "LLMError"]

providers/llm/nearai_provider.py

@@ -0,0 +1,142 @@
+"""
+NEAR AI Cloud LLM provider.
+
+OpenAI-compatible chat completions over NEAR AI Cloud TEE inference.
+"""
+
+import os
+import time
+from typing import Any, Dict, Iterator, List, Optional
+
+from providers.llm.base import LLMError, LLMProvider, LLMResponse
+from providers.registry import ProviderType, registry
+
+
+class NearAIProvider(LLMProvider):
+    """NEAR AI Cloud provider via OpenAI-compatible REST API."""
+
+    DEFAULT_BASE_URL = "https://cloud-api.near.ai/v1"
+    DEFAULT_MODEL = "zai-org/GLM-5.1-FP8"
+
+    def __init__(self, config: Dict[str, Any] = None) -> None:
+        super().__init__(config)
+        self.api_key = self._resolve_api_key()
+        self.base_url = self._resolve_base_url()
+        self.default_model = self._config.get("default_model", self.DEFAULT_MODEL)
+
+    def _resolve_api_key(self) -> str:
+        key = self._config.get("api_key", "")
+        if key and not key.startswith("${"):
+            return key
+        return os.getenv("NEARAI_API_KEY", "")
+
+    def _resolve_base_url(self) -> str:
+        base_url = (
+            self._config.get("base_url")
+            or self._config.get("baseUrl")
+            or os.getenv("NEARAI_BASE_URL", "")
+            or self.DEFAULT_BASE_URL
+        )
+        return base_url.rstrip("/")
+
+    def _chat_url(self) -> str:
+        return f"{self.base_url}/chat/completions"
+
+    def _build_messages(
+        self,
+        messages: List[Dict[str, str]],
+        system_prompt: Optional[str],
+    ) -> List[Dict[str, str]]:
+        full_messages: List[Dict[str, str]] = []
+        if system_prompt:
+            full_messages.append({"role": "system", "content": system_prompt})
+
+        for message in messages:
+            role = message.get("role", "user")
+            if role == "developer":
+                role = "system"
+            full_messages.append({
+                "role": role,
+                "content": message.get("content", ""),
+            })
+        return full_messages
+
+    def generate(
+        self,
+        messages: List[Dict[str, str]],
+        system_prompt: Optional[str] = None,
+        model: Optional[str] = None,
+        **kwargs,
+    ) -> LLMResponse:
+        try:
+            import requests  # type: ignore
+        except ImportError:
+            raise LLMError("nearai", "requests library not installed")
+
+        model = model or self.default_model
+        max_tokens = kwargs.get("max_tokens", kwargs.get("max_completion_tokens", 512))
+        timeout = kwargs.get("timeout", 30)
+
+        payload: Dict[str, Any] = {
+            "model": model,
+            "messages": self._build_messages(messages, system_prompt),
+            "max_tokens": max_tokens,
+        }
+
+        if "temperature" in kwargs:
+            payload["temperature"] = kwargs["temperature"]
+        if "tools" in kwargs:
+            payload["tools"] = kwargs["tools"]
+        if "tool_choice" in kwargs:
+            payload["tool_choice"] = kwargs["tool_choice"]
+
+        start = time.time()
+        try:
+            resp = requests.post(
+                self._chat_url(),
+                headers={
+                    "Authorization": f"Bearer {self.api_key}",
+                    "Content-Type": "application/json",
+                },
+                json=payload,
+                timeout=timeout,
+            )
+            resp.raise_for_status()
+        except Exception as exc:
+            raise LLMError("nearai", f"API request failed: {exc}") from exc
+
+        data = resp.json()
+        latency_ms = (time.time() - start) * 1000
+        choice = data["choices"][0]
+
+        return LLMResponse(
+            content=choice["message"]["content"],
+            model=model,
+            provider="nearai",
+            usage=data.get("usage", {}),
+            latency_ms=latency_ms,
+            finish_reason=choice.get("finish_reason", "stop"),
+            raw_response=data,
+        )
+
+    def generate_stream(
+        self,
+        messages: List[Dict[str, str]],
+        system_prompt: Optional[str] = None,
+        model: Optional[str] = None,
+        **kwargs,
+    ) -> Iterator[str]:
+        response = self.generate(messages, system_prompt, model, **kwargs)
+        yield response.content
+
+    def is_available(self) -> bool:
+        return bool(self.api_key)
+
+    def get_info(self) -> Dict[str, Any]:
+        info = super().get_info()
+        info["name"] = self._config.get("name", "NEAR AI Cloud")
+        info["base_url"] = self.base_url
+        return info
+
+
+registry.register(ProviderType.LLM, "nearai", NearAIProvider)

routes/admin.py

@@ -601,6 +601,16 @@ def list_clients():
 
 _OPENCLAW_CONFIG_PATH = Path('/app/runtime/openclaw.json')
 
+# Static admin list sampled from GET https://cloud-api.near.ai/v1/model/list on 2026-05-21.
+_NEARAI_MODELS = [
+    {'id': 'zai-org/GLM-5.1-FP8', 'name': 'GLM 5.1', 'contextWindow': 202752},
+    {'id': 'Qwen/Qwen3.6-35B-A3B-FP8', 'name': 'Qwen 3.6 35B A3B FP8', 'contextWindow': 262144},
+    {'id': 'Qwen/Qwen3.5-122B-A10B', 'name': 'Qwen3.5 122B A10B', 'contextWindow': 131072},
+    {'id': 'Qwen/Qwen3-VL-30B-A3B-Instruct', 'name': 'Qwen3-VL 30B A3B Instruct', 'contextWindow': 256000},
+    {'id': 'google/gemma-4-31B-it', 'name': 'Gemma 4 31B Instruct', 'contextWindow': 262144},
+    {'id': 'openai/gpt-oss-120b', 'name': 'GPT OSS 120B', 'contextWindow': 131000},
+]
+
 # Known providers and their config shape
 _AI_PROVIDERS = {
     'mx': {
@@ -664,6 +674,13 @@ def list_clients():
             {'id': 'llama-4-maverick-17b-128e-instruct', 'name': 'Llama 4 Maverick', 'contextWindow': 131072},
         ],
     },
+    'nearai': {
+        'name': 'NEAR AI Cloud',
+        'envKey': 'NEARAI_API_KEY',
+        'baseUrl': 'https://cloud-api.near.ai/v1',
+        'api': 'openai-completions',
+        'models': _NEARAI_MODELS,
+    },
     'google': {
         'name': 'Google Gemini',
         'envKey': 'GEMINI_API_KEY',

setup-config.js

@@ -22,6 +22,16 @@ const PORT = getKey("PORT") || "5001";
 const token = crypto.randomBytes(24).toString("hex");
 const secret = crypto.randomBytes(32).toString("hex");
 
+// Sampled from GET https://cloud-api.near.ai/v1/model/list on 2026-05-21.
+const NEARAI_MODELS = [
+  { id: "zai-org/GLM-5.1-FP8", name: "GLM 5.1", contextWindow: 202752 },
+  { id: "Qwen/Qwen3.6-35B-A3B-FP8", name: "Qwen 3.6 35B A3B FP8", contextWindow: 262144 },
+  { id: "Qwen/Qwen3.5-122B-A10B", name: "Qwen3.5 122B A10B", contextWindow: 131072 },
+  { id: "Qwen/Qwen3-VL-30B-A3B-Instruct", name: "Qwen3-VL 30B A3B Instruct", contextWindow: 256000 },
+  { id: "google/gemma-4-31B-it", name: "Gemma 4 31B Instruct", contextWindow: 262144 },
+  { id: "openai/gpt-oss-120b", name: "GPT OSS 120B", contextWindow: 131000 },
+];
+
 // ---------------------------------------------------------------------------
 // 1. Write openclaw.json (nested gateway/agents format for v2026.3.2+)
 // ---------------------------------------------------------------------------
@@ -67,6 +77,20 @@ const openclawConfig = {
   // Plugins configured by individual plugin installers
 };
 
+if (getKey("NEARAI_API_KEY")) {
+  openclawConfig.models = {
+    mode: "merge",
+    providers: {
+      nearai: {
+        baseUrl: "https://cloud-api.near.ai/v1",
+        api: "openai-completions",
+        apiKey: "${NEARAI_API_KEY}",
+        models: NEARAI_MODELS,
+      },
+    },
+  };
+}
+
 fs.mkdirSync("openclaw-data/workspace", { recursive: true });
 fs.writeFileSync(
   "openclaw-data/openclaw.json",
@@ -95,6 +119,7 @@ const envLines = [
 const envKeyList = [
   "ANTHROPIC_API_KEY", "OPENAI_API_KEY", "GEMINI_API_KEY",
   "OPENROUTER_API_KEY", "MISTRAL_API_KEY", "XAI_API_KEY", "ZAI_API_KEY",
+  "NEARAI_API_KEY",
   "CEREBRAS_API_KEY", "TOGETHER_API_KEY", "HF_TOKEN",
   "MOONSHOT_API_KEY", "KIMI_API_KEY", "MINIMAX_API_KEY", "QIANFAN_API_KEY",
   "MODELSTUDIO_API_KEY", "XIAOMI_API_KEY", "VOLCANO_ENGINE_API_KEY",
@@ -135,6 +160,7 @@ const providerMap = {
   MISTRAL_API_KEY: "mistral",
   XAI_API_KEY: "xai",
   ZAI_API_KEY: "zai",
+  NEARAI_API_KEY: "nearai",
   CEREBRAS_API_KEY: "cerebras",
   TOGETHER_API_KEY: "together",
   HF_TOKEN: "huggingface",

src/admin.html

@@ -1756,6 +1756,7 @@
     llm:[
       {id:'gateway',name:'OpenClaw Gateway',detail:'GLM-5-turbo via OpenClaw. Full tools, memory, canvas.'},
       {id:'zai',name:'Z.AI Direct',detail:'GLM-5-turbo / GLM-5.1. No agent tools.'},
+      {id:'nearai',name:'NEAR AI Cloud',detail:'TEE inference via OpenClaw. OpenAI-compatible.'},
       {id:'groq',name:'Groq',detail:'Llama 4 Scout / Qwen3-32b. Fast inference.'},
       {id:'minimax',name:'MiniMax',detail:'M2.7-highspeed. Custom provider "mx".'},
       {id:'kimi',name:'Kimi',detail:'K2.5 / Moonshot AI.'},