ByteDance Doubao 2.1 Pro Officially Released — Costs Slashed by 80% to Match Claude Opus 4.7, Seedance 2.5 Previews 30-Second Video Generation, WeChat “XiaoWei” Launches AI Assistant Beta

6.24 On June 24, at the FORCE Power Conference, ByteDance officially released the Doubao 2.1 Pro large model, with core metrics that shook the industry: programming capabilities on par with Anthropic's Claude Opus 4.7, and overall usage costs reduced by nearly 80% compared to the Claude Opus 4.6–4.8 series. At the same time, the company previewed that its video generation model Seedance 2.5 will officially launch in early July,

On June 24, at the FORCE Power Conference, ByteDance officially released the Doubao 2.1 Pro large model, with core metrics that shook the industry: programming capabilities on par with Anthropic's Claude Opus 4.7, and overall usage costs reduced by nearly 80% compared to the Claude Opus 4.6–4.8 series. At the same time, the company previewed that its video generation model Seedance 2.5 will officially launch in early July, supporting single-video generation of up to 30 seconds. WeChat's AI assistant "XiaoWei" simultaneously began small-scale beta testing, marking ByteDance's AI ecosystem's full-scale penetration from the "model layer" to the "application layer."

"80% Cost Slash": The "Value-for-Money Revolution" of Chinese Large Models

Doubao 2.1 Pro's pricing strategy continues the "ultimate cost-performance" approach of Chinese AI vendors:

DimensionDoubao 2.1 ProClaude Opus 4.7Gap
Programming CapabilityOn parBaselineTied
Overall Cost80% reductionBaseline1/5 the cost
Token Call Volume10x+ YoY growth (as of June)—Explosive growth
Video GenerationSeedance 2.5 (30 sec / early July)No direct competitorDifferentiated


This "80% cost slash" is not simply a price war, but a systematic cost reduction based on ByteDance's self-developed chips (in collaboration with Pingtou Ge), Volcano Engine compute optimization, and model architecture innovation. As of June 2026, Doubao's token call volume has grown over 10x year-over-year, validating the effectiveness of the "low price for scale" strategy.

WeChat "XiaoWei": An Entry Point Revolution from "Mini Programs" to "AI Skills"

WeChat has rolled out its biggest update ever—the AI assistant "XiaoWei" has entered small-scale beta testing, with core features including:

  • Base model: Tencent WeLM (not Doubao—highlighting WeChat ecosystem's independence)

  • Feature coverage: Voice/text command execution for conversations, document reading, transfers, and Moments management

  • Mini-program transformation: Mini-programs transition from "standalone apps" to "Skills" that can be invoked by AI, supporting services like food ordering and ride-hailing.

This means WeChat's 1.3 billion monthly active users will gradually enter the era of "AI-native interaction"—users no longer open individual mini-programs one by one, but simply tell "XiaoWei," "Order me the highest-rated spicy hotpot nearby," and the AI automatically invokes the Meituan mini-program to complete the order.

Conclusion

ByteDance's "dual-engine" strategy (Doubao models + Seedance video + WeChat XiaoWei) is forging the strongest closed loop in China's AI ecosystem. Doubao 2.1 Pro's "80% cost slash" is not an isolated event, but a systematic advancement of the Chinese AI industry's "open-source + low-cost + ecosystem" strategy—alongside DeepSeek-V4-Pro's 75% price cut, Alibaba's Qwen3.7-Max ranking first in 12 domestic evaluations, and Zhipu's GLM-5.2 topping CodeArena globally—all collectively mounting a "cost restructuring" assault on the Western closed-source, high-margin business model.

WeChat "XiaoWei's" beta launch carries even greater strategic weight: while US AI assistants (Siri, Alexa, Google Assistant) are still struggling for "entry-point" status, WeChat already possesses a ready-made super-app gateway. Mini-programs evolving from "apps" to "skills" signals that China's internet ecosystem is shifting from "people finding services" to "AI matching services"—this is not a technology upgrade, but a revolution in interaction paradigms. However, challenges remain: the technical gap between WeLM and Doubao, controversies over WeChat ecosystem's closed nature, and privacy boundaries around AI-invoked mini-programs will all face scrutiny once "XiaoWei" is fully rolled out.