AI资讯 - 大模型产业动态

技术前沿

GLM Coding Plan Real Quota Revealed: How Many Tokens Do Lite, Pro & Max Actually Give You?

Deep dive into Zhipu GLM Coding Plan's real token-based billing: 5-hour/weekly/monthly limits across Lite, Pro, Max tiers, China vs international pricing, and a cost comparison against API pay-per-use billing using real production data (198:1 input-output ratio).

2026-07-05 2 次阅读

模型发布

Three Chinese AI Models to Watch in July 2026: DeepSeek V4, Kimi K3, and GLM-5.3

DeepSeek V4 full release, Kimi K3 with linear attention, and GLM-5.3 with multimodal — three Chinese LLMs converging on July 2026 with breakthroughs in coding, agent engineering, and long-context understanding.

2026-07-04 2 次阅读

观点评论

DeepSeek-V4 Peak-Valley Pricing: AI Computing Enters the Electricity Era

DeepSeek V4 launches mid-July 2026 with a peak-off-peak pricing model — API prices double during peak hours. It's more than a price adjustment; it's the inflection point where AI computing shifts from flat-rate supply to refined operations.

2026-07-01 1 次阅读