Error during streaming: 500: Failed to obtain VQD token

[KiKaraage]

Hi, I just installed this service today. I tried it with Elelem as the wrapper (Fedora 42).

• It detected the model well, I got all the 5 model options.
• Though I can't get it to work due to failure to obtain VQD tokens (see logs)
• Another issue is, seems like the models are outdated? Currently Duck.ai has
  • GPT 4o-mini
  • o4-mini (replacing o3-mini)
  • Claude 3.5 Haiku (replacing Claude 3 Haiku)
  • Llama 4 Scout (replacing Llama 3.3 70b Instruct Turbo)
  • Mistral Small 3 (replacing Mixtral 8x7b)

Logs from the terminal:

keyless-gpt-wrapper-api on  master via 🐍 v3.13.5 
❯ python server.py
2025-07-06 18:28:15,554 - WARNING - No TikTok session ID provided. TTS functionality will be disabled.
2025-07-06 18:28:15,554 - INFO - TikTok TTS functionality disabled - set TIKTOK_SESSION_ID environment in your docker-compose or docker run command 
 or pass --session-id argument to enable if running locally
INFO:     Started server process [16124]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:1337 (Press CTRL+C to quit)
INFO:     127.0.0.1:57432 - "OPTIONS /v1/chat/completions HTTP/1.1" 200 OK
2025-07-06 18:33:44,079 - INFO - Received chat completion request for conversation 1b12b353-cbaa-4b5c-82ef-3a748b624471
2025-07-06 18:33:44,080 - INFO - Request: {'model': 'keyless-gpt-o3-mini', 'messages': [{'role': 'user', 'content': 'What is Tailscale?', 'audio': None}], 'conversation_id': None, 'temperature': 1.0, 'top_p': 1.0, 'n': 1, 'stream': True, 'stop': None, 'max_tokens': None, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'logit_bias': None, 'user': None, 'modalities': None, 'audio': None}
INFO:     127.0.0.1:57440 - "POST /v1/chat/completions HTTP/1.1" 200 OK
2025-07-06 18:33:44,485 - INFO - HTTP Request: GET https://duckduckgo.com/country.json "HTTP/1.1 200 OK"
2025-07-06 18:33:44,564 - INFO - HTTP Request: GET https://duckduckgo.com/duckchat/v1/status "HTTP/1.1 200 OK"
2025-07-06 18:33:44,564 - INFO - Fetched new x-vqd-4 token: 
2025-07-06 18:33:44,565 - ERROR - Error during streaming: 500: Failed to obtain VQD token
2025-07-06 18:34:50,935 - INFO - Received chat completion request for conversation f5f972d0-4b8b-48ed-a63d-773ea518f213
2025-07-06 18:34:50,935 - INFO - Request: {'model': 'keyless-gpt-4o-mini', 'messages': [{'role': 'user', 'content': 'Tell me something about Kubernetes', 'audio': None}], 'conversation_id': None, 'temperature': 1.0, 'top_p': 1.0, 'n': 1, 'stream': True, 'stop': None, 'max_tokens': None, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'logit_bias': None, 'user': None, 'modalities': None, 'audio': None}
INFO:     127.0.0.1:44548 - "POST /v1/chat/completions HTTP/1.1" 200 OK
2025-07-06 18:34:51,402 - INFO - HTTP Request: GET https://duckduckgo.com/country.json "HTTP/1.1 200 OK"
2025-07-06 18:34:51,486 - INFO - HTTP Request: GET https://duckduckgo.com/duckchat/v1/status "HTTP/1.1 200 OK"
2025-07-06 18:34:51,486 - INFO - Fetched new x-vqd-4 token: 
2025-07-06 18:34:51,486 - ERROR - Error during streaming: 500: Failed to obtain VQD token
2025-07-06 18:35:04,103 - INFO - Received chat completion request for conversation 236d62a9-b57a-47d4-a5ee-332f1626b38b
2025-07-06 18:35:04,103 - INFO - Request: {'model': 'keyless-meta-Llama-3.3-70B-Instruct-Turbo', 'messages': [{'role': 'user', 'content': 'Tell me something about Github', 'audio': None}], 'conversation_id': None, 'temperature': 1.0, 'top_p': 1.0, 'n': 1, 'stream': True, 'stop': None, 'max_tokens': None, 'presence_penalty': 0.0, 'frequency_penalty': 0.0, 'logit_bias': None, 'user': None, 'modalities': None, 'audio': None}
INFO:     127.0.0.1:46764 - "POST /v1/chat/completions HTTP/1.1" 200 OK
2025-07-06 18:35:04,232 - INFO - HTTP Request: GET https://duckduckgo.com/country.json "HTTP/1.1 200 OK"
2025-07-06 18:35:04,309 - INFO - HTTP Request: GET https://duckduckgo.com/duckchat/v1/status "HTTP/1.1 200 OK"
2025-07-06 18:35:04,310 - INFO - Fetched new x-vqd-4 token: 
2025-07-06 18:35:04,311 - ERROR - Error during streaming: 500: Failed to obtain VQD token

As comparison, I tried GPT 4o-mini on duck.ai with the same question and it worked as usual.

Hope you can fix this!

[beamer2k23]

I encountered the same issue two weeks ago. Possibly duck.ai has changed the data contained in the website source code? The VQD hash value is still present in the chat endpoint, but the VQD value itself is not.
I have tried a workaround by fetching the VQD (whatever that means) from the search endpoint, but there seem to be dependencies between VQD value and VQD hash value which I am not aware of.
Here are some lines of code for further inspiration to fix the issue:

url = "https://duckduckgo.com/"
params = {"q": dummy_query}
headers = {
"User-Agent": user_agent,
"Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,/;q=0.8",
"Referer": "https://duckduckgo.com/",
"Origin": "https://duckduckgo.com",
}

resp = requests.get(url, headers=headers, params=params, timeout=5)
resp.raise_for_status()
html = resp.text

match = re.search(r"vqd=([\d-]+)&", html)
vqd_token = match.group(1)

[beamer2k23]

@callbacked, do you have any idea how to proceed with this issue?