Gunjo · Business Intelligence for the AI Era
← Sticker Wall MODEL · DETAIL

Zhipu Open-Source Model API Token Economy: 27x Revenue Growth Validates Self-Sustaining Business Model

1) Cloud API charged by token usage, offering free trial credits with separate pricing for enterprise tiers and private

MODEL

Key Fields

FIELD STAMPS
IndustryAI / LLM
RegionChina
ScaleGiant
ChannelOnline

📌 Background

In 2026, the commercialization of open-source large language models entered deep waters; open-source weights are no longer free, and leading vendors are shifting to API-based monetization. According to Zhipu's 2026 interim report, first-half revenue reached 954 million RMB, a year-on-year increase of 399.7%. Among this, open platform and API business revenue hit 825 million RMB, up 2,735.7% year-on-year, accounting for 86.5% of total revenue (based on company financial reporting). Token call volume grew by over 40 times compared to the beginning of the year, and the average API selling price increased by approximately 101%.

👤 Target Customers

Application developers and enterprise IT departments, paying on a per-API-call/token usage basis.

💰 Revenue Streams

1) Cloud API charged by token usage, offering free trial credits with separate pricing for enterprise tiers and private deployments; 2) Elastic scaling: tiered overage fees charged when usage exceeds quotas, with additional capacity fees for reserved dedicated capacity; 3) Key enterprise dedicated deployment: project-based separate settlement for deployment and integration fees for government and enterprise clients requiring private deployment or integration with internal platforms.

🧮 Cost Structure

Model training computing power, GPU inference clusters, R&D teams, open-source community operations, and compliance costs.

🛡️ Moat

Continuous foundation model iteration capability, open-source community ecosystem scale, and gross margin headroom driven by token-level pricing and inference cost optimization.

🔑 Keys to Success

  • Continuously open-sourcing new models to maintain technological influence
  • Optimizing inference costs to sustain token pricing advantages
  • Building developer toolchains and industry-specific solutions

⚠️ Risks

  • API price wars compressing profit margins
  • Enterprise private deployment of open-source models slowing public API demand growth

🏢 Cases

  • Zhipu AI

📊 SWOT Analysis

Strengths

  • Open-source strategy lowers market education costs and rapidly drives developer adoption
  • High-speed growth in API revenue validates customer willingness to pay

Weaknesses

  • Severe commoditization of large model APIs with intense price competition
  • High computing power investment relies on external supply chains

Opportunities

  • Growth in enterprise-grade AI-native applications driving API consumption
  • Inference efficiency improvements can further expand gross margins

Threats

  • Price pressure from low-cost APIs such as DeepSeek
  • Some major enterprise clients shifting to private deployments, reducing cloud-based calls