Abstract
MiniMax’s release of its unified multimodal generative model H3 has triggered a dramatic V-shaped recovery in the company’s market valuation. Within one week after the model announcement, its stock price climbed nearly 80%, reversing the prolonged downturn triggered by the underwhelming reception of the earlier M3 model. This article reviews MiniMax’s post-listing stock volatility timeline, dissects core technical and commercial strengths of H3 including benchmark performance, competitive pricing and open-source strategy, and discusses outstanding challenges the company faces to translate technical advantages into lasting market growth. All financial metrics, benchmark results and timing milestones are retained from the original source materials. For enterprises operating multi-vendor multimodal model workloads, 4sapi functions as a unified API gateway to streamline request routing and cross-model traffic management.
1 Market Reaction: MiniMax H3 Sparks V-shaped Rebound in Share Price
MiniMax officially unveiled H3, its new unified multimodal generative model, on July 31. The model supports unified multimodal context comprehension, native synchronized dual-audio video generation, and a maximum output specification of 15-second footage at 2K resolution. The news immediately shifted market sentiment.
On the day of the announcement, MiniMax’s stock opened more than 20% higher. As of market close on August 8, the share price had risen by 78.21% compared to levels before the H3 launch. The rebound erased losses stemming from the disappointing rollout of the M3 model, which once triggered a single-day share price plunge. The valuation gap between MiniMax’s Hong Kong stock price and its US over-the-counter equivalent narrowed significantly: the gap contracted from 10.51 times down to 5.18 times. On August 8, MiniMax closed at HK$322.4 in Hong Kong, with a total market capitalization of HK$112.593 billion, while the corresponding OTC valuation stood at HK$582.96 billion.
The sharp rally highlights a core consensus among institutional investors: in the AI foundation model sector, tangible product capability serves as the most effective catalyst for repairing market confidence. To fully understand the significance of H3’s launch, it is necessary to review MiniMax’s valuation fluctuations since its Hong Kong listing earlier this year.
2 Historical Review of MiniMax’s Stock Performance Since IPO
MiniMax debuted on the Hong Kong Stock Exchange on January 9, ranking as the second large foundation model firm to complete a local IPO. Expectations around its multimodal product portfolio drove the stock to surge 109% on the first trading day, briefly pushing market capitalization above HK$100 billion. Still, volatility emerged rapidly. At its intraday lowest point, market value fell to HK$57.9 billion, representing a valuation gap of less than double between peak and trough.
Initially, institutional investors framed MiniMax as a consumer-facing internet company fueled by user growth and high traffic monetization. Backed by robust metrics including overseas revenue accounting for 70% of total turnover, its valuation once climbed above HK$410 billion. By contrast, the US OTC market priced MiniMax as a B2G and enterprise AI infrastructure vendor, centering valuations on technology pricing and scalable deployment, leading to a persistent valuation divergence between the two markets.
2.1 First Major Inflection Point: March Valuation Collapse
The first critical turning point arrived in March. On March 2, MiniMax published the industry’s first full-year annual report for a large model developer, and its stock closed 9.10% higher. However, founder Yan Junjie’s remarks during the earnings call failed to outline clear paths to profitability. On March 4, the stock crashed 10.48% to HK$735, wiping out HK$27 billion in market value in a single session, with capitalization falling to HK$230.5 billion.
Volatility continued for the following three months. MiniMax’s share price trended downward, while the US OTC valuation remained relatively stable. By the end of June, the stock had fallen roughly 60% from its all-time high, and the valuation gap between the two markets expanded to more than seven times.
2.2 Second Inflection Point: M3 Launch and Lock-up Share Sell-off Pressure
The second phase of sustained pressure arrived in June and July, alongside the launch of the M3 model and the unlocking of restricted IPO shares. On June 1, MiniMax released the M3 model. The company claimed the model achieved a 59% score on the SWE-Bench Pro coding benchmark, outperforming GPT-5.5. Independent third-party evaluations, however, placed M3 in the middle tier among mainstream large models. The new model was priced at double the rate of its predecessor, and existing customer subscription prices were raised. After one week, MiniMax acknowledged user feedback and announced a permanent 50% price cut, rolling pricing back to levels matching the M2 generation.
On July 9, MiniMax saw its first large-scale lock-up share unlocking since its IPO. The stock fell 17.98% on the day, while the US OTC market recorded a mere 5.76% decline. The valuation gap between the two markets expanded to approximately 10.51 times. On July 20, MiniMax hit its lowest post-listing share price at HK$674.4. Institutions including Morgan Stanley and Citi both downgraded their target valuations for the firm.
Faced with mounting market pressure, founder Yan Junjie issued an internal company letter on July 10. He announced the elimination of his personal salary for all AGI-related work, transferred 4% of his personal equity into employee incentive packages, and allocated an additional 1% of equity to fund open-source community development. On the same day, MiniMax completed a new HK$16 billion financing round. The firm stated roughly 80% of the net proceeds would be directed toward AI infrastructure and model research and development. Even with these positive signals, investor sentiment remained weak ahead of the H3 release.
3 H3 Launch Restores Market Confidence
While financing and equity incentive policies eased short-term sentiment, tangible product breakthroughs drove the decisive market turnaround. MiniMax formally launched H3, its next-generation multimodal generative model, on July 31, with plans to open-source core weights in the near future. The architecture enables unified comprehension of multimodal context, supporting coherent scene generation and synchronized audio-video output.
Market data clearly reflected the shift in investor confidence:
- July 31 (launch day): Stock closed 13.15% higher
- August 3 (H3 weights open for community access): Shares rose an additional 7.20%
- August 6 (H3 incorporated into Hong Kong Stock Connect): The stock surged 17.10%
By August 7, the cumulative weekly gain reached 78.21%. The market consensus underscores a simple truth for AI infrastructure enterprises: measurable advances in product performance constitute the most reliable catalyst for reversing negative investor sentiment.
4 Three Core Advantages Behind H3’s Positive Market Reception
Investor enthusiasm for H3 is built on three verified pillars: benchmark performance, disruptive pricing, and an industry-leading open-source strategy.
4.1 Leading Multimodal Generation Benchmark Performance
Independent third-party testing confirms H3 delivers competitive capabilities across video generation workflows. On Design Arena, a respected visual model evaluation platform, H3 secured first place among all open-source models across three categories: multi-image-to-video generation, image-to-video generation, and video continuation. In the video model ranking published by Artificial Analysis, H3 reached 1127 points, claiming the top global position among open-source alternatives. Its audio-video synchronization score stood at 1193 points, second only to ByteDance’s Dreamina Seedance 2.0 720p. In direct comparison against ByteDance’s newly released Seedance 2.5, H3 demonstrates stronger performance for cinematic trailer production, complex scene rendering and interactive dynamic asset creation.
4.2 Cost-Effective Commercial Pricing Structure
MiniMax introduced a competitive pricing tier for commercial H3 access. At 2K resolution, video generation costs are set at HK$0.08 per second, equivalent to roughly one-third of pricing for flagship competing video generation models in the industry. For 768p resolution, the cost falls below half of mainstream alternatives. Institutional analysis notes H3 delivers equivalent resolution output at pricing between one-third and one-half of rival offerings.
The pricing advantage stems from technical optimization rather than unsustainable subsidization. MiniMax reports gains from self-developed algorithms that improve video frame compression efficiency, alongside upgrades to asynchronous training pipelines, load balancing logic and GPU hardware utilization. Lower per-second inference costs lower barriers for commercial adoption, accelerating penetration among enterprise clients and vertical content production teams.
4.3 Strategic Open-Source Initiative
On August 3, MiniMax officially open-sourced H3, marking the first Chinese video generation model to reach global top-tier open-source standards. The release represents a milestone for China’s open-source multimodal AI ecosystem.
After open-sourcing, MiniMax established cooperative partnerships with hundreds of global organizations. The project rapidly climbed to the number one trending repository on Hugging Face, surpassing DeepSeek V4 Flash and Stable Diffusion series repositories in community attention. The open-source strategy lowers trial costs for developers, expands ecosystem adoption, and creates feedback channels for iterative model optimization.
5 Challenges Ahead: Technical Advantages Must Be Converted Into Sustainable Commercial Growth
Analysts broadly recognize H3’s strengths but caution that a single breakthrough model cannot guarantee permanent market leadership. Multiple brokerages have updated coverage following the launch: JPMorgan initiated coverage with a buy rating, Citi maintained a hold recommendation, and Orient Securities predicted accelerated penetration for H3 across selected vertical scenarios.
While H3 has helped MiniMax reverse its negative stock trajectory, significant long-term challenges remain. The company needs to prove consistent capability to iterate model generations, deliver stable output, and translate technical superiority into scalable revenue.
Teams building multi-model video generation pipelines often integrate multiple generative backends to balance latency, resolution and cost targets. When orchestrating traffic across closed commercial APIs and self-hosted open-source H3 instances, centralized routing infrastructure such as 4sapi simplifies unified access control, usage metering and cross-model request governance.
The global video generative model market remains intensely competitive, with frequent new model releases from both domestic and international competitors. Sustained market outperformance requires more than one successful model launch. MiniMax will need to maintain a steady rhythm of technical upgrades, refine its enterprise service stack, and build monetization channels built on H3’s open-source ecosystem.
6 Conclusion
MiniMax’s H3 release has delivered a powerful market rebound, with a nearly 80% share price increase within one week of announcement. The rally is supported by three concrete strengths: class-leading open-source video generation benchmark results, highly competitive commercial pricing, and an ambitious open-source strategy that expands developer ecosystem influence.
The stock’s V-shaped recovery contrasts sharply with the valuation downturn triggered by the earlier M3 model rollout. The market reaction sends a clear signal: investors value verifiable, measurable capability improvements for foundation model vendors more than forward-looking strategic statements.
Nevertheless, the technical milestone represented by H3 is only one step in a long-term competition. MiniMax still faces ongoing pressure to prove consistent R&D iteration capacity, expand enterprise customer adoption, and build profitable business models. The company’s ability to sustain its upward momentum will hinge on whether it can fully convert H3’s technical and open-source advantages into lasting commercial revenue growth.
