Tencent's Hunyuan Hy4 Preview Launches: Targeting Kimi and GLM for Productivity Tasks
AlchainHust · x · 2026-08-28
Tencent released the Hunyuan Hy4 preview model, scoring 1633 on the third-party Arena leaderboard, rivaling Grok-4.6 and Claude Fable 5. Official benchmarks show it closely matches GPT 5.6 Sol and Kimi K3. With 770B total parameters and 1M context, it targets code and office productivity tasks. Exclusively available on WorkBuddy for a two-week free trial, it offers aggressive pricing at 6 RMB input and 18 RMB output per million tokens.
More from Models
- DeepSeek Flash is one tweak away from Opus-class performance, says Bindu Reddy — bindureddy · 2026-08-28
- Ant launches Ling-3.0-flash-Fin for finance workflows; free OpenRouter access — WarInspiron · 2026-08-28
- Alibaba's Qwen3.8-Flash activates only 6B of 125B params, cuts costs by 90% — 大模型之路 · 2026-08-28
- Fal to release open-source weights for H3 Max model — listopalafoto · 2026-08-28
- Rumor: Google internally testing Gemini 3.8 Flash — Able-Line2683 · 2026-08-28
- Critique of GLM-3.5-Flash: Needs steering, not on par with GPT-4o — kevinkern · 2026-08-28