Hi everyone, I've made several updates to the Quinn - The Alibaba cloud native pipeline since the initial submission in Samsar: The prompt to visualize anything framework.
Now you can almost run the entire-stack using an Alibaba Cloud pay-as-you-go token, with minimal external fallback for Lip-sync, sound-effects and TTS (To be integrated via Alibaba Cloud token as well soon) Please see the updated model matrix available via various types of token plans-
- Inference: Qwen 3.8 Max
- Text to image: Wan 2.7 Pro
- Image to video: Happy Horse 1.1 I2V
Pay-As-You-Go Token
Models: Qwen 3.8 Max, Qwen Image 3.0 Pro, Wan 2.7 Pro, Happy Horse 1.1 I2V
Token Plan
Models: Qwen 3.8 Max, Wan 2.7 Pro, Happy Horse 1.1 I2V
Subscription Plan Token
Models: Qwen 3.8 Max, Wan 2.7 Pro, Happy Horse 1.1 I2V
Coding Plan Token
Models: Qwen 3.8 Max, Wan 2.7 Pro, Happy Horse 1.1 I2V
Final availability follows the models enabled for the connected Alibaba Cloud account.
Reliability Improvements
- Repairs overlong generated dialogue while preserving valid lines.
- Normalizes JSON
typetypos before validation. - Returns structured errors for malformed JSON and invalid value types.
- Uses extended token budgets, timeouts, and controlled retries for complex Qwen requests.
Check out the latest code here
Log in or sign up for Devpost to join the conversation.