Doubao AI Takes a Quantum Leap with Version 2.0

In a significant move for China's AI landscape, Volcano Engine has unveiled Doubao Large Model 2.0 - marking substantial improvements in performance, affordability and accessibility.

Power Meets Affordability

The standout feature? A dramatic 90% reduction in inference costs compared to industry leaders. Since its initial release, daily Tokens usage has skyrocketed over 500 times - testament to growing adoption.

"We've rebuilt the architecture from the ground up," explains a Volcano spokesperson. "This isn't just incremental improvement - it's transformational efficiency."

Four Models, Endless Possibilities

The update introduces specialized versions:

• Pro: The powerhouse for complex reasoning and agent tasks
• Lite: Balanced performance that outshoots version 1.8 while using fewer resources
• Mini: Lightning-fast responses ideal for cost-sensitive applications
• Code: Developer-focused with enhanced IDE integration

Seeing is Understanding

The model's visual comprehension now competes globally:

• Leads Gemini3pro in spatial/motion/video understanding benchmarks
• Excels at real-time video stream analysis
• Can guide users through complex physical activities like billiards or fitness routines

"Imagine an AI personal trainer watching your form," suggests ML researcher Li Wei. "That's the level of motion perception we're achieving."

Smarter Conversations, Better Decisions

The upgraded LLM capabilities shine in professional contexts:

• Outperforms GPT5.2 in SuperGPQA evaluations
• Tops HealthBench rankings
• Matches top models in scientific domains

The enhanced Agent functionality handles multi-step workflows seamlessly - from research to conclusions - making it invaluable for customer service automation.

Developers Rejoice

The newly opened API access comes with attractive pricing:

• First-month access starts at just ¥8
• Coding Plan package optimized for programming tasks
• TRAE integration slashes development time dramatically

"We built 'AI Temple Fair' with just five prompt rounds," shares early adopter Zhang Tao. "The efficiency gains are unreal."

Key Points:

✅ 90% cost reduction makes AI more accessible than ever
✅ Four specialized models cater to diverse application needs
✅ World-class visual understanding enables innovative use cases
✅ Enhanced coding support accelerates developer productivity
✅ New API access opens doors for enterprise integration