Zhipu AI ARR Hits $1 Billion, Growing 15x in Six Months
N.R. Finch
As of July 2026, Zhipu's ARR has reached $1 billion, growing 15× since January — a milestone that took Anthropic 15 months but Zhipu just five, as the coding-first AI strategy reshapes China's large-model commercialization timeline.
How fast is $1 billion ARR, really?
According to multiple independent sources cited by智能涌现 (SmartSurge), Zhipu's ARR hit $1 billion by July 2026. Zhipu declined to comment.
From January to July 2026, ARR grew 15×. Investors had previously expected the $1–1.5 billion mark by year-end at the earliest.
In plain terms = the market gave Zhipu a full-year runway; it finished in half.
How does this compare to Anthropic?
Going from $100 million to $1 billion ARR, Anthropic took about 15 months; Zhipu took five.
Anthropic's Claude Code pushed its valuation past the $1 trillion mark in May 2026, overtaking OpenAI.
This means → once the coding route proves commercially viable, the valuation ceiling rises fast. Anthropic set the template; Zhipu is running the same playbook.
Why the coding-first bet?
Zhipu was among the first Chinese large-model companies to go all-in on coding. In early 2025 it narrowed R&D focus to model coding capability.
Founder Tang Jie framed the thesis as "Coding and Reasoning — a model capability that can co-evolve with agents."
GLM-5.2, released open-source in June 2026, matched or exceeded Claude Opus 4.8 and GPT-5.5 on multiple core benchmarks.
This reflects a broader signal: open-source models can now trade blows with closed-source leaders, and coding ability is the lever closing the gap.
Prices up 83% — and usage still grew 400%?
In Q1 2026, GLM's API prices rose a cumulative ~83%. Overseas subscription pricing approached Claude Code levels.
Yet call volume still grew roughly 400%.
Zhipu defines large-model commercial value as intelligence ceiling × token consumption scale. In plain terms = as long as the model is strong enough, users don't leave when prices rise — they use it more.
Rivals are accelerating — can Zhipu hold on?
In June, MiniMax released M3, focused on coding and agent capabilities. On July 16, Moonshot AI launched K3 with 2.8 trillion parameters, reportedly second only to Claude Fable 5 and GPT-5.6 Sol in general intelligence.
In the U.S., GPT-5.6's release and the ChatGPT-CodeX merger show OpenAI closing the gap with Anthropic on coding.
Video generation is scaling in parallel: Seedance 2.0 is pulling in nearly ¥1 billion in monthly revenue; Goldman Sachs projects Kling AI's ARR could reach $1 billion by year-end.
This means → Zhipu's $1 billion ARR is a milestone, not a moat. The more certain the coding track becomes, the more competitors it attracts — whether Zhipu can sustain this growth rate is the next key test.
Content is for reference only, not financial advice.