🔥36氪•Stalecollected in 1m
MiniMax Plans IPO on Shanghai STAR Market
💡One of China's leading LLM startups is moving toward a public listing, signaling a major shift in the AI capital market.
⚡ 30-Second TL;DR
What Changed
MiniMax has formally resolved to explore an IPO on the STAR Market.
Why It Matters
This move signals the growing maturity of China's top-tier AI startups as they seek domestic capital to fuel large-scale model training and infrastructure expansion. It may set a precedent for other Chinese LLM companies aiming for local public listings.
What To Do Next
Monitor MiniMax's public filings for insights into their infrastructure spending and compute resource allocation strategies.
Who should care:Founders & Product Leaders
Key Points
- •MiniMax has formally resolved to explore an IPO on the STAR Market.
- •The company has signed a guidance agreement with professional consultants.
- •The potential issuance of RMB-denominated shares is subject to regulatory approval and market conditions.
🧠 Deep Insight
Web-grounded analysis with 17 cited sources.
🔑 Enhanced Key Takeaways
- •MiniMax successfully completed an initial public offering (IPO) on the Hong Kong Stock Exchange in January 2026, raising approximately $619 million.
- •Following its Hong Kong debut, MiniMax's shares surged by approximately 400% by late May 2026, pushing its market capitalization to around $33 billion.
- •The company's annual recurring revenue (ARR) surpassed $300 million by late May 2026, a figure that more than doubled in the preceding two months, partly due to the success of its M2.7 model.
- •MiniMax is actively competing with Zhipu AI to become the first major Chinese large language model company to achieve a domestic A-share listing on the STAR Market.
- •Founded in December 2021 by former SenseTime researchers Yan Junjie and Zhou Yucong, MiniMax initially received funding from gaming company MiHoYo.
📊 Competitor Analysis▸ Show
| Feature/Metric | MiniMax | Zhipu AI | Moonshot AI | DeepSeek | OpenAI | Anthropic |
|---|---|---|---|---|---|---|
| Key Models | M-series (M1, M2, M2.5, M2.7), ABAB 6.5, MiniMax-01 (Text-01, VL-01), Hailuo AI (Video-01), Speech-02 | GLM series (e.g., GLM 5.1) | Kimi K2 Thinking, Kimi K2.6 | DeepSeek V4 Pro, R1 model | GPT series (e.g., GPT-5, GPT-5.5) | Claude series (e.g., Claude Sonnet 4.5, Claude Opus 4.6, Claude 4.7) |
| Core Capabilities | Multimodal (text, audio, image, video, music), long context, agentic, coding, self-evolving models | LLM, tool calls, agentic | Long-context LLM (up to 200,000 Chinese characters), multimodal (Kimi-V1-Vision-Preview), agentic (300 parallel sub-agents) | AGI focus, math, coding, multimodal, cost-effective | Generative AI, LLM, multimodal | Generative AI, LLM, multimodal |
| Benchmarks (Selected) | M2.7: QI 49.6, M2 ahead of Gemini 2.5 Pro | GLM 5.1: strong on linguistic depth/cultural nuance | Kimi K2 Thinking outperformed GPT-5, Claude Sonnet 4.5 on some benchmarks. Kimi K2.6 scored 54 on Artificial Analysis's Intelligence Index | DeepSeek V4 Pro: 1.6 trillion parameters, MIT licensed, Codeforces rating 3206 | GPT-5.5 at 60 on Artificial Analysis's Intelligence Index | Claude Opus 4.7 at 57 on Artificial Analysis's Intelligence Index |
| Pricing (Examples) | M2.5: $0.15/M input tokens, $1.20/M output tokens. M2.5 Lightning: $0.30/M input, $2.40/M output. | (Not explicitly found for specific models) | Kimi K2.6 API: ~$0.73/M input, ~$3.49/M output (OpenRouter) | (Not explicitly found for specific models) | (Generally higher than Chinese counterparts) | (Generally higher than Chinese counterparts, e.g., Claude Opus 4.6 20x cost of MiniMax M2.5) |
| Noteworthy | Dual A+H listing, strong international revenue, founded by ex-SenseTime. | Competing for A-share listing, on U.S. Entity List since Jan 2025. | Strong in long-context processing. | Research-driven, cost-effective solutions. | Leading proprietary models. | Leading proprietary models. |
🛠️ Technical Deep Dive
- MiniMax develops a suite of multimodal foundation models capable of understanding, generating, and integrating text, audio, image, video, and music.
- Key model series include the M-series (M1, M2, M2.5, M2.7), ABAB 6.5 (mixture of experts), and MiniMax-01 (Text-01, VL-01).
- The M-series models are engineered for efficiency, with M1 focusing on long context and M2/M2.1 on high-throughput agent and coding workflows.
- MiniMax M2.7 boasts a 204,800 token context window and is noted as the world's first self-evolving model, capable of improving its own training process.
- The M2.5 and M2.5 Lightning models utilize transformer-based designs with a focus on balancing capability with computational requirements, offering performance comparable to more expensive systems at a fraction of the cost.
- The M2 series employs an "interleaved thinking" protocol, appending the full thinking history directly into the conversation context for enhanced agentic capabilities.
- MiniMax is developing the M3 series, which will feature a new sparse attention mechanism designed to yield up to 15.6 times faster decoding speed for million-token contexts using a custom sub-quadratic framework.
- Consumer applications built on these models include AI character apps Talkie and Xingye, and the video-generation service Hailuo AI.
🔮 Future ImplicationsAI analysis grounded in cited sources
MiniMax's dual listing strategy will significantly enhance its access to capital and liquidity.
Listing on both the Hong Kong and Shanghai STAR markets allows MiniMax to tap into a broader investor base, including mainland Chinese investors, which can provide substantial funding for its capital-intensive AI development and expansion.
The competition between MiniMax and Zhipu AI for the first A-share LLM listing will accelerate innovation and market consolidation in China's AI sector.
The race to be the first major Chinese LLM company on the A-share market creates intense pressure for both companies to demonstrate superior technology, commercialization, and financial performance, potentially leading to faster product cycles and strategic moves.
MiniMax's secondary listing on the STAR Market will strengthen its ties with domestic policy support and industrial ecosystems, potentially boosting its presence in the Chinese mainland.
A domestic listing can align MiniMax more closely with national strategic technology initiatives and provide opportunities for securing benchmark projects within China, especially as a company that currently derives a significant portion of its revenue from overseas.
⏳ Timeline
2021-12
MiniMax founded by former SenseTime researchers Yan Junjie and Zhou Yucong.
2022-10
Launched Glow, an AI character application.
2023-06
Talkie, an AI character app, launched for international markets.
2024-03
Alibaba Group led a $600 million financing round, valuing MiniMax at $2.5 billion, and the company launched Hailuo AI.
2025-07
MiniMax secured nearly $300 million in a funding round, reaching a post-money valuation of approximately $4 billion.
2026-01
MiniMax completed its initial public offering (IPO) on the Hong Kong Stock Exchange, raising approximately $619 million.
2026-05
MiniMax filed an A-share listing guidance report with the Shanghai Securities Regulatory Bureau, initiating its secondary listing process on the STAR Market.
📎 Sources (17)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 36氪 ↗