clarity-ai 6.5.0 → 6.5.1

package/README.md

@@ -103,16 +103,10 @@ clarity /bash ls -la
 | Gemini 2.0 Flash | Google | 32K (fast) |
 | DeepSeek R1 Free | OpenRouter | 128K |
 
-## Provider Comparison
+## Architecture
 
-| Provider | Free Tier | Streaming | Priority |
-|---|---|---|---|
-| Groq | ✓ | ✓ | 1 (fastest) |
-| Google Gemini | ✓ | ✓ | 2 |
-| HuggingFace (Clarity Flash) | Needs HF_TOKEN | ✓ | 3 |
-| DeepSeek | Cheap | ✓ | 4 |
-| OpenRouter | ✓ | ✓ | 5 |
+Clarity Flash 14B is served via `Clarity-main` HF Space — a streaming inference proxy at `*.hf.space/v1/chat/completions`. No API keys needed.
 
 ## License
 
-MIT
+Proprietary — all rights reserved. See [LICENSE](LICENSE).

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "clarity-ai",
-  "version": "6.5.0",
+  "version": "6.5.1",
   "description": "CLARITY — terminal AI agent with Clarity Flash 14B",
   "type": "module",
   "bin": {

package/requirements.txt

@@ -1,4 +1,3 @@
-gradio>=4
-requests>=2
-uvicorn>=0.20
 fastapi>=0.100
+httpx>=0.27
+uvicorn>=0.20

package/space_app.py

@@ -1,28 +1,31 @@
-import os, json, requests
-import gradio as gr
+import os, json, asyncio
+import httpx
 from fastapi import FastAPI, Request
 from fastapi.responses import StreamingResponse, JSONResponse
-import uvicorn
+from contextlib import asynccontextmanager
 
 HF_TOKEN = os.environ.get('HF_TOKEN', '')
-MODEL = 'Universal-618/Clarity-flash-weights'
+MODEL = os.environ.get('HF_MODEL', 'Universal-618/Clarity-flash-weights')
 API = f'https://api-inference.huggingface.co/models/{MODEL}/v1/chat/completions'
 HEADERS = {'Authorization': f'Bearer {HF_TOKEN}', 'Content-Type': 'application/json'}
 
 app = FastAPI()
 
 @app.post('/v1/chat/completions')
-async def chat_completions(request: Request):
+async def proxy(request: Request):
     body = await request.json()
     body['model'] = MODEL
     stream = body.get('stream', True)
     body['stream'] = stream
-    resp = requests.post(API, headers=HEADERS, json=body, stream=True)
-    if resp.status_code != 200:
-        return JSONResponse({'error': resp.text}, status_code=resp.status_code)
-    if stream:
-        return StreamingResponse(resp.iter_lines(), media_type='text/event-stream', headers={'Cache-Control': 'no-cache', 'Connection': 'keep-alive', 'X-Accel-Buffering': 'no'})
-    return resp.json()
+    async with httpx.AsyncClient(timeout=120) as client:
+        resp = await client.post(API, headers=HEADERS, json=body)
+        if resp.status_code != 200:
+            return JSONResponse({'error': await resp.aread()}, status_code=resp.status_code)
+        if stream:
+            return StreamingResponse(resp.aiter_lines(), media_type='text/event-stream', headers={
+                'Cache-Control': 'no-cache', 'Connection': 'keep-alive', 'X-Accel-Buffering': 'no'
+            })
+        return resp.json()
 
 @app.get('/main-data/{path:path}')
 async def main_data(path: str):
@@ -33,20 +36,10 @@ async def main_data(path: str):
     with open(fp) as f:
         return JSONResponse(json.load(f))
 
-def chat_ui(message, history):
-    history = history or []
-    msgs = [{'role': 'user' if i % 2 == 0 else 'assistant', 'content': m} for i, m in enumerate(history + [message])]
-    body = {'model': MODEL, 'messages': msgs, 'stream': False, 'max_tokens': 4096}
-    resp = requests.post(API, headers=HEADERS, json=body)
-    if resp.status_code == 200:
-        return resp.json()['choices'][0]['message']['content']
-    return f'Error: {resp.status_code}'
-
-with gr.Blocks(title='CLARITY Flash', css='footer{display:none!important}') as ui:
-    gr.Markdown('# CLARITY Flash 14B')
-    gr.ChatInterface(chat_ui, title='Chat')
-
-app = gr.mount_gradio_app(app, ui, path='/')
+@app.get('/')
+async def root():
+    return {'status': 'ok', 'model': MODEL, 'docs': '/v1/chat/completions'}
 
 if __name__ == '__main__':
+    import uvicorn
     uvicorn.run(app, host='0.0.0.0', port=7860)

package/src/providers/index.js

@@ -2,17 +2,26 @@ import { getKey } from '../config/keys.js';
 import { streamResponse } from './streaming.js';
 import { parseErrorResponse } from './errors.js';
 
+const SPACES = [
+  'https://universal-618-clarity-main.hf.space',
+  'https://universal-618-clarity-2.hf.space',
+  'https://universal-618-clarity-3.hf.space',
+  'https://universal-618-clarity-4.hf.space',
+  'https://universal-618-clarity-5.hf.space',
+  'https://universal-618-clarity-6.hf.space',
+];
+
 const PROVIDERS = {
   huggingface: {
-    endpoint: 'https://Universal-618-Clarity-main.hf.space/v1/chat/completions',
+    endpoints: SPACES.map(s => s + '/v1/chat/completions'),
     name: 'huggingface',
   },
   groq: {
-    endpoint: 'https://api.groq.com/openai/v1/chat/completions',
+    endpoints: ['https://api.groq.com/openai/v1/chat/completions'],
     name: 'groq',
   },
   openrouter: {
-    endpoint: 'https://openrouter.ai/api/v1/chat/completions',
+    endpoints: ['https://openrouter.ai/api/v1/chat/completions'],
     name: 'openrouter',
   },
 };
@@ -40,17 +49,26 @@ export async function* callAI(providerName, model, messages, options = {}) {
     body.tool_choice = 'auto';
   }
 
-  let endpoint = provider.endpoint;
   const extraHeaders = {};
   if (providerName === 'openrouter') {
     extraHeaders['HTTP-Referer'] = 'https://clarity-ai.local';
     extraHeaders['X-Title'] = 'CLARITY AI';
   }
 
-  const stream = streamResponse(endpoint, body, key || 'none', extraHeaders, options.signal);
-  for await (const event of stream) {
-    yield event;
+  let lastErr;
+  for (const endpoint of provider.endpoints) {
+    try {
+      const stream = streamResponse(endpoint, body, key || 'none', extraHeaders, options.signal);
+      for await (const event of stream) {
+        yield event;
+      }
+      return;
+    } catch (err) {
+      lastErr = err;
+      continue;
+    }
   }
+  throw lastErr;
 }
 
 export async function callAISync(providerName, model, messages, options = {}) {