Tencent Launches Hy4flash: 1/20 the Cost of Hy4 With Faster Agent Execution
APPSO · wechat · 2026-11-16
Tencent released Hy4flash, a 224B-parameter model with 12B activated, supporting multimodal input and 1M context, aimed at delivering Hy4's agent, long-horizon and productivity capabilities at lower cost. Pricing is ¥0.5/M input and ¥2/M output tokens—about a tenth of Hy4—with per-task cost on WorkIndex at 1/13 of Hy4preview, 1/20 of Hy4, and 1/8 of DeepSeek-V4.1-Flash (max tier, peak pricing). Free daily quota in WorkBuddy from Nov 16–30.
APPSO's hands-on: with identical prompts and tool permissions, Hy4flash finished a deep-research task 10 minutes faster than DeepSeek V4.1 Flash; an interactive 3D sailing-ship build took Hy4flash 30 minutes (¥0.23) vs DeepSeek's 90 minutes (¥1.21), though DeepSeek produced richer detail. Across 10 coding and office tasks, final quality was comparable but Hy4flash delivered first drafts noticeably faster. Verdict: Hy4flash for high-frequency office, research and prototyping; Hy4 for complex long-horizon engineering. Also tested: PDF-to-Markdown/Word/PPT conversion, multi-agent expert teams, and a three.js 3D RPG.
More from Models
- Dev praises Gemini 4 carbon as his favorite model despite 20-minute waits — willcb · 2026-11-17
- Qwen 3.8 27B vs Claude Fable 5.2: a $0 local model (65 vs 20 on AA Index) and which jobs it can take over — alex_verem · 2026-11-17
- Blogger finds Astra skips steps on research tasks while carbon dutifully finishes them — lxfater · 2026-11-17
- 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