KNOWLEDGE ATLAS, a leading global large language model company, issued an apology letter on February 21st regarding its GLM Coding Plan, and announced corresponding handling and compensation measures.
The company stated that the recent service revision primarily involved three mistakes: insufficient transparency of the rules, an overly slow grayscale rollout schedule for GLM-5, and a poorly designed upgrade mechanism for existing users.
KNOWLEDGE ATLAS explained that following the release of GLM-5, traffic significantly exceeded expectations. The company's infrastructure expansion pace could not keep up, necessitating a phased release of GLM-5 access in the order of Max, Pro, and Lite tiers. Currently, Max tier users have full access. While Pro tier users now have access, they may experience rate limiting during peak hours due to high cluster loads. Lite tier users will be gradually granted access through a grayscale release during off-peak hours after the holiday period. For affected Lite and Pro users, the company supports self-initiated refund applications.
Notably, on February 20th, KNOWLEDGE ATLAS (02513.HK) saw its share price close up against the market trend, surging 42.72% in a single day to HKD 725, setting a new record high. The company's total market capitalization reached HKD 323.2 billion. Within 43 days of its listing, the share price has accumulated a gain of 524% compared to its IPO price of HKD 116.2.
On the same day, Kuaishou (01024.HK) closed down 2.78% with a market cap of HKD 289.4 billion; JD.com (09618.HK) closed down nearly 2% with a market cap of HKD 330.9 billion; Trip.com (09961.HK) closed down nearly 1% with a market cap of HKD 296.7 billion; and Baidu Group (09888.HK) closed down 6.25% with a market cap of HKD 354.8 billion. This indicates that KNOWLEDGE ATLAS has joined the top tier of Hong Kong-listed TMT stocks by market capitalization.
KNOWLEDGE ATLAS officially launched its new flagship model, GLM-5, on February 12th. It demonstrates an average performance improvement of over 20% in programming and development scenarios compared to the previous generation, with real-world coding experience approaching the level of Claude Opus 4.5. The model achieved the best performance in the open-source domain across three Agent evaluations: BrowseComp, MCP-Atlas, and τ2-Bench.
Following the release of GLM-5, due to overwhelming demand, KNOWLEDGE ATLAS simultaneously increased the prices of its GLM Coding Plan packages the next day. Prices rose by 30% for the China region and over 100% for the international version, making it the first AI-native company in China to raise prices for commercialized large model services. The new packages sold out immediately upon launch, setting a new industry record for a domestic AI programming model's paid plan selling out.
According to a recent research report from J.P. Morgan, KNOWLEDGE ATLAS's API pricing serves as a signal of leading-edge capabilities, with its real-time market clearing signals containing multi-dimensional information about the model's prowess. This price increase is seen as the clearest reflection of enhanced model capabilities.
The report pointed out that KNOWLEDGE ATLAS has reached a significant inflection point, particularly in its global API business. The releases of GLM-4.5/4.6/4.7, coupled with a clear strategic shift towards agent systems, tool-augmented reasoning, and developer-facing infrastructure, indicate the company is aligning its technology roadmap with the dimensions that increasingly define global cutting-edge capabilities. The institution expects that as recognition of GLM-5 grows among developers, its usage rate will increase rapidly, especially within coding-centric workflows which typically exhibit higher willingness to pay and usage intensity.
Furthermore, as a recent listing from just this January, a substantial 60% of KNOWLEDGE ATLAS AI's shares allocated for strategic investors are still within their lock-up period, resulting in an extremely scarce float of actually tradable shares. Against the backdrop of strong capital inflows into the AI sector, the minimal selling pressure has amplified share price volatility to an extreme.
Comments