MiniMax
D
Dolphin Research

Policy and competition squeeze: will China LLMs pivot to Ascend?

Last Fri., DeepSeek pushed a 4.1 update; while only a minor point release, the performance uplift is clear. Coupled with same-day headlines that DS plans to use Ascend for its next-gen pretraining and a ban issued by the U.S. Dept. of Homeland Security, the recent DS newsflow has meaningful implications for the sector and the supply chain.

$Z.AI(02513.HK) $MINIMAX-W(00100.HK) core takeaways compiled. Core takeaways:

1. How is the model? V4.1-Flash-1017 is not a 'China beats Opus' narrative...

10:42C a New chat Hi,I'm DeepSeek. How can I help you today? Message DeepSeek 88
Intelligence Index vs. Cost per Task Artificial Analysis Intelligence Index · We
DeepSeek DeepSeek Kimi GLM GPT Claude Benchmarks V4.1-Flash- V4.1-Flash K3 5.3 5
Ascend 950DT vs Nvidia: A Generation Behind per Chip, Compensated by Scale (950D+4
D
Dolphin Research

Bearish calls overdone: MiniMax isn't that dire

China’s independent model developer $MINIMAX-W(00100.HK) released its H1 results after the HK close on Aug 26. The company has been gradually losing ground in 2026.

Using Mar and Jun as key checkpoints, MiniMax has moved almost one-way down. Its market cap fell from $55bn to $13.5bn.

With the actual numbers now out, is it really that bad? Let’s take a closer look (for the earnings call Trans, click here): 1) Revenue: in a breakout year for coding...

咖 参 MINIMAX ANIM
Minimax revenues AI Products Al enterprise service Revenues --Total revenues yoy
Minimax revenues mix % of AI products % of enterprise service % barcn Minimax Ha
Revenues by region 1,400 earcn 90% 1,200 DolphinRes kearc 78% 70% 73% 73% 73% 72+14
D
Dolphin Research

MiniMax’s sky‑high valuation: bubble or the future?

In the business-model analysis 'Deep Dive into MiniMax and Zhipu: Are Foundation Models, Compute Intensity and Financing Stamina the Decider?', Dolphin Research argued that amid high costs and rapid iteration, the foundation-model industry has turned into a brutal test of financing endurance.

But the true core of that money magnetism rests on two fundamentals for model companies: 1) model capability; 2) product and commercial execution.

This piece focuses on how $MINIMAX-W(00100.HK) performs on those two fronts...

MIN MI MIN MM IN MIN iMA M INIMA MINIMAX MINIMAXMINIMAX MIN M MIN N MA MI IM MIN
Minimax Type Version Release Version Gap API Pricing Text Model abab 1 4/22 / ab
End to End Response Time to First Answer Model Output Price ($/mn LLM Model Prov
MiniMax M2.7 by minimax ea arch Artificial Analysis Intelligence Index: 49.6 $0.+36
D
Dolphin Research

LLM losses widened 360%, MiniMax still a 'darling'?

Old guard bows out, new contender steps in. Today, HK's 'hot pick' and China's independent model developer $MINIMAX-W(00100.HK) released its FY2025 report card. However, vs. a share price up 4.5x in two months, Dolphin Research's estimated Q4 print looks softer.

Let's dig in: I) Revenue — consumer slowed, enterprise accelerated; is the mix improving? A 'C down, B up' pattern points to a healthier mix.

1) Revenue growth eased: FY2025 revenue totaled $79 mn (+~160% YoY), still surging. With 3Q revenue already disclosed, we can infer...

MINIMAX
Minimax revenues AI Products Al enterprise service Revenues Total revenues yoy%
Minimax revenues mix % of Al products % of enterprise service % 100% Research rc
Minimax Type Version Release Version Gap APIPricing abab 1 4/22 / abab 5.5 11/23+7
D
Dolphin Research

MiniMax (Trans): Doubling down on coding, productivity, and multimodal use cases

Below is Dolphin Research's MiniMax FY25 earnings call Trans.

Section 1 covers the detailed discussion.

1.1 outlines management's core messages.

1) Language model. For model iteration, M2/M2.1 were updated in 2025, and M2.5 was released in Feb 2026.

Breakthroughs centered on coding, tool invocation, and office use cases.

b) Key metrics. M2.5 set a new SWE-bench record in coding.

Token efficiency improved 37% vs M2.1...

图片 1,共 1 张
D
Dolphin Research

Minimax and Zhipu: The Ultimate Test for Large Models, Computing Power Intensity, and Funding Endurance?

A deep dive into Minimax and Zhipu: are large language models, compute intensity, and funding stamina the decisive edge? In early 2026, near the third anniversary of ChatGPT's launch, two leading Chinese AI model startups $MINIMAX-W(00100.HK) and $Z.AI(02513.HK) listed at around $6bn valuations.

Both stocks rallied after listing, driving a broad-based surge across AI application names. Yet a close read of their financials shows seemingly endless losses. On one hand, a brutal global price war in LLMs is turning the stack increasingly into a commodity...

智谱 普·AI MINIMAX MiniMax 智谱
$100mn 176 163 160 152 157 159 147 148 144 139 136 141 126 118 122 117 115 114 1
Revenues comparison Minimax Zhipu USD/0,000 6,000 DolphinResearch DolphinResearc
Gross margin comparison Minimax 一■ Zhipu Costs and opex distribution % % 70% 64.+9