Mechanical dragonfly test: Hy4 Flash uses 4x fewer tokens but DeepSeek build is better
TencentHunyuan · x · 2026-11-17
User ItsmeAjayKV had two flash-tier models one-shot a mechanical flying dragonfly with the same prompt, no harness, no tool calls.
- DeepSeek V4.1 Flash 1017 (68,304 thinking + 14,708 answer tokens, 83,012 output total) produced the better, more mechanical result
- Hy4 Flash used only 20,661 output tokens (1/4 of DeepSeek) and got the job done, but the build was more basic and fell short of expectations
- Takeaway: token consumption differs massively for the same task, and DeepSeek currently wins on quality
Related event: Tencent Hy4 Flash Benchmarked Against DeepSeek V4.1 Flash(2 posts)→
More from Models
- Reuters exclusive: China taps US AI models via distillation to train homegrown rivals — ns123abc · 2026-11-17
- IFM releases K2-Horizon: six fully open LLMs up to 375B MoE, trained on ~10T synthetic tokens — IFM · 2026-11-17
- Optimization researcher: Astra proved with one prompt what we couldn't — bremen79 · 2026-11-17
- Gemini's AI Overview caught fabricating brake rotor dimensions for a Fisker Ocean query — AWTom · 2026-11-17
- Dev Slams Claude for Building a Content Repository With No Way to Add Content — dejanseo · 2026-11-17
- You can't bypass AI detectors — a random word list still scores 100% AI on Pangram — rubenhassid · 2026-11-17