Tencent's new Hy4 Flash AI model slashes API costs by 90% for massive 1M-token context
Original headline: Tencent Hy4 Flash — hy.tencent.ai
More in AI and machine learning news
Tencent Cloud has announced Hy4 Flash, a new variant of its Hy4 AI model specifically engineered for cost efficiency. The model maintains the core Hy4 architecture with 224 billion total parameters (including 64 billion 'Engram' parameters) but uses only 12 billion active parameters per token, while keeping native multimodality and a massive 1 million token context window.
The key innovation is dramatically reduced API pricing: input processing costs drop from $0.834 to just $0.070 per million tokens, output generation from $2.501 to $0.281, and cached input from $0.042 to $0.003—representing savings of over 90% in some categories. This positions Hy4 Flash as a practical option for developers needing long-context AI capabilities without the premium pricing, potentially accelerating adoption of multimillion-token applications in everyday productivity tools.
Read the original at hy.tencent.ai
Satirical commentary on real tech news, summarised with AI. Read the original source for the facts. How ZeroDaily works
