Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Tencent releases and open-sources Hy4 flash model family

AI summary

Tencent Hunyuan released Hy4 flash, a natively multimodal model with a 1M-token context window, and made the standard, FP8 and Base variants available for download on major model platforms. Tencent describes it as its most cost-efficient Hy model, with 12B active parameters per token and substantially lower API pricing than Hy4. Independent office-workflow testing found Hy4 flash produced usable documents and web pages with relatively low usage, while most benchmark and cost comparisons remain vendor-defined.

In-depth insight

The release gives developers an open-weight, multimodal model family positioned for frequent productivity and agent workloads at lower inference cost, with downloadable variants for local deployment and fine-tuning.

November 16, 2026

Introducing Hy4 flash: Ultimate cost-efficiency for everyday productivity

We’re introducing Hy4 flash, our most cost-effective Hy model to date—featuring 224B total parameters (including 64B Engram parameters) with just 12B active parameters per token. Like Hy4, it is natively multimodal with a 1M-token context window, while offering significantly lower API pricing.

API pricing (per million tokens)Cached inputInputOutput
Hy4 flash$0.003$0.070$0.281
Hy4$0.042$0.834$2.501

Across training, evaluation, and inference, we aggressively optimized for cost efficiency while preserving intelligence:

  • Pre-training: Hy4 flash integrates hybrid attention, LatentMoE, and Engram into its architecture—reducing KV cache size to just 1/24 of Hy3 and 1/6 of Hy4, with extensive scaling ladder experiments ensuring performance meets expectations.
  • Post-training: Self-distillation from Hy4 establishes strong general intelligence, while reinforcement learning cuts average end-to-end output tokens by 50%.
  • Evaluation: We built WorkIndex, an internal benchmark grounded in typical WorkBuddy tasks, and paired it with human evaluation—focusing on deliverable quality, single-turn output length, and task turnaround time rather than public benchmark scores detached from everyday work.
  • Inference: Combining megakernels, 4-bit quantization, and DFly pushes decoding speed past 150 tokens/s, enabling low-cost serving at scale.

#Tencent Hunyuan#Hy4 flash#Hy4 flash-FP8#Hy4 flash Base#WorkBuddy#Tencent Cloud TokenHub#Yuanbao#WeCom#QQ#ima#Marvis#DeepSeek V4.1 Flash

Original · 3 min readRead the original
Group chat

No comments yet. Start the conversation.