🐯虎嗅•Stalecollected in 23m
LLMs Tested for Lazy Responses
#llm-benchmarks#user-complaints#prompt-engineeringdoubaodeepseekdoubaoyuanbaoqianwenwenxin-yiyan
💡Real tests reveal which LLMs slack—optimize your prompts now
⚡ 30-Second TL;DR
What Changed
Doubao initially generated only 2/10 consumer rights posters, needed prompting for rest.
Why It Matters
Exposes UX gaps in free LLMs, pushing providers to balance cost vs depth; users must refine prompts.
What To Do Next
Benchmark your LLM with 10-poster gen and Forbes sorting tasks for laziness.
Who should care:Developers & AI Engineers
Key Points
- •Doubao initially generated only 2/10 consumer rights posters, needed prompting for rest.
- •Doubao excels in classifying Forbes billionaire list by continent/country.
- •DeepSeek, Qianwen provide partial oil price data; Yuanbao has factual errors.
- •Models self-admit or dodge 'laziest' ranking; Doubao owns up most.
🧠 Deep Insight
Background and context from public sources — not the original article. 8 sources cited.
🔑 Enhanced Key Takeaways
- •Doubao leads the Chinese AI chatbot market with 172 million monthly active users as of 2025, holding a 19% advantage over DeepSeek's 145 million and a 5x lead over Yuanbao.
- •Yuanbao significantly boosted its user base to over 10 million daily actives after integrating the DeepSeek model, improving stability in tasks like search and content recommendations.
- •ByteDance's algorithmic recommendation expertise gives Doubao superior search accuracy compared to competitors like Yuanbao.
- •Doubao's 'everything app' strategy integrates multimodal features and Douyin social capabilities, capturing 40% of users who migrated from DeepSeek by October 2025.
📊 Competitor Analysis▸ Show
| Metric | Doubao | DeepSeek | Yuanbao |
|---|---|---|---|
| Monthly Active Users | 172M (2025) | 145M (2025) | ~35M (inferred) |
| Multimodal Capabilities | Text, image, video, voice | Primarily text, limited | Improved post-DeepSeek integration |
| Social Integration | Deep (Douyin) | Minimal | WeChat ecosystem |
| Pricing | Ultra-low | Low (5x higher than Doubao) | Promotional |
| Technical Reputation | Strong UX | Excellent math/logic | Vertical scenarios (education, images) |
| Server Stability | Generally stable | Frequent traffic issues | Stable post-integration |
🛠️ Technical Deep Dive
- •DeepSeek-V3: 600-671B parameters (37B active), enhanced Mixture of Experts (MoE) architecture pre-trained on 15 trillion tokens, excels in code generation with AIME 2025 score of 89.3.
- •DeepSeek-R1: 671B parameters (37B active), supports chain-of-thought reasoning and multi-token prediction.
- •DeepSeek V3.2: 685B parameters, S-tier benchmarks including GPQA Diamond 79.9 and Chatbot Arena 1421 rating.
🔮 Future ImplicationsAI analysis grounded in cited sources
Doubao will maintain >15% market lead through 2026
Its multimodal and social integrations have already captured 40% of DeepSeek migrants, per 2025 data.
DeepSeek model integrations will proliferate in Chinese apps
Yuanbao's 10x DAU growth post-DeepSeek integration demonstrates its stabilizing effect on competitors.
Cost controls will persist, prioritizing UX over raw benchmarks
Doubao's ultra-low pricing and mass-market focus overtook DeepSeek's technical lead by late 2025.
⏳ Timeline
2024-12
DeepSeek V3 released, establishing strong coding benchmarks.
2025-01
DeepSeek achieves #1 peak user position in China.
2025-04
Doubao regains #1 position with multimodal and social features.
2025-10
Doubao records 11.4M App Store downloads, 5x DeepSeek's.
2025-12
DeepSeek V3 full release or DeepSeek-R1 launch with advanced reasoning.
2026-01
Yuanbao integrates DeepSeek, achieving 10M+ DAU.
📎 Sources (8)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
- dataglobehub.com — Doubao Statistics and Insights
- slashdot.org — Deepseek vs Doubao
- recodechinaai.substack.com — Chinas Three Kingdoms in AI Bytedance
- alphamatch.ai — Open Source LLM Comparison Blog 2026
- till-freitag.com — Open Source LLM Comparison
- vertu.com — Open Source LLM Leaderboard 2026 Rankings Benchmarks the Best Models Right Now
- chinatalk.media — Chinese AI Rings in the Year of the
- hackernoon.com — Choosing an LLM in 2026 the Practical Comparison Table Specs Cost Latency Compatibility
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 虎嗅 ↗
This is a summary, not the original. Read the source, or get the weekly briefing.
Weekly AI briefing
One email a week. Unsubscribe anytime.