SHELFCRITTER AI Digest

COMPANY

MiniMax

companytopic-notechina

Overview

MiniMax is a Chinese AI frontier lab known for the MiniMax-M2 sparse-activation MoE family and a pattern of public-market fundraising that contrasts with the US private-round playbook. Following a $619M Hong Kong debut in January 2026, MiniMax is now layering a mainland A-share listing on top of that float.

Timeline

  • 2026-06-01-AI-Digest — MiniMax filed a listing guidance report with the Shanghai Securities Regulatory Bureau on 2026-05-29, kicking off an A-share IPO process with CITIC Securities as guidance institution (Commerce & Finance Law Offices and EY Hua Ming on counsel/audit). Mainland listing follows the $619M Hong Kong debut in January 2026 (HK$165, +109% day one; shares now ~HK$840, market cap ~HK$263.5B). Pairs with Vast‘s unicorn round as evidence the Chinese frontier-lab cohort (excluding still-private DeepSeek) is converging on capital-markets fundraising.

  • 2026-06-02-AI-Digest — MiniMax announces MiniMax M3 with a new MiniMax Sparse Attention architecture claiming ~1/20th compute at 1M tokens, 9× faster input and 15× faster generation vs dense attention at long context, trained on 100T interleaved multimodal tokens, with weights set to drop to Hugging Face and GitHub within 10 days. Vendor-published benchmarks: SWE-Bench Pro 59% (ahead of GPT-5.5 and Gemini 3.1 Pro, just behind Opus 4.7) and BrowseComp 83.5 (beats Opus 4.7’s 79.3). Unaudited, but the first credible sparse-attention efficiency numbers at long context from the open-weight cohort.

  • 2026-07-15-AI-Digest — MiniMax surfaces as one of the five Chinese labs sweeping the top-6 OpenRouter slots (with Tencent, Xiaomi, DeepSeek, Z.ai) in TechCrunch’s Chinese-open-weight-distribution-majority story; Chinese-open-weight downloads take 41% of Hugging Face spring share. No fresh MiniMax product action today; log as distribution-share thread reference.

  • 2026-06-13-AI-Digest — Two simultaneous arXiv drops from MiniMax. (1) MaxProof (arXiv:2606.13473) — MiniMax M3 trains proof generation, verification, and critique-conditioned repair into one model, then runs a population search with tournament selection across candidate proofs at test time, reportedly scoring 35/42 on IMO 2025 and 36/42 on USAMO 2026 — a population-level test-time-scaling result on mathematical reasoning. (2) MiniMax Sparse Attention (arXiv:2606.13392) — blockwise sparse attention on top of GQA selecting relevant blocks per query, with 28.4× compute reduction and 14.2× prefill / 7.6× decode speedup on H800. Coordinated mathematical-reasoning + long-context push from one lab in a single drop.

  • 2026-08-05-AI-Digest — MiniMax releases MiniMax H3, the first open-weights video model to top an AI video ranking — a Chinese-lab open-weights video release named on the White House exemption-carve-out list alongside DeepSeek, Alibaba‘s Qwen, and Moonshot AI‘s Kimi K3. Simon Willison‘s hands-on run of PipeNetwork/minimax-h3-mlx on an M5 Pro MacBook Pro reports ~115 GB weights and ~45 minutes to generate a 15-second video from a text prompt (with caveats about audio-prompting failure modes). Narrow read: open-weights video generation just became runnable on a laptop, and the top-of-leaderboard placement means the open ecosystem is now inside striking distance of the closed video-gen frontier. Structural read the corpus carries: the “Chinese lab ships open weights, Simon Willison writes the practitioner cookbook within 48 hours” pattern is now a reliable release-observation loop for open Chinese frontier drops — same shape as recent Alibaba / Qwen and Moonshot AI / Kimi K3 releases. 30-day watch: whether MiniMax H3 appears in any leaderboard managed OUTSIDE China (Chatbot Arena, Artificial Analysis) to confirm the ranking survives independent evaluation.

  • 2026-09-02-AI-Digest — MiniMax co-founder Yeyi Yun proposes 1% of global GDP as the AGI milestone at the Goldman Sachs Asia Leadership Conference in Hong Kong on 2026-09-01 (Bloomberg) — measuring AGI by whether AI systems can self-generate the equivalent of 1% of global GDP — roughly ~$1.1T on today’s ~$110T global GDP — rather than by benchmark scores. Narrow read the digest carries: do NOT frame this as a novel category of AGI definition — economic-output framings already exist: the Microsoft–OpenAI arrangement has carried a $100B-profit AGI trigger since 2023, and OpenAI’s own charter uses “highly autonomous systems that outperform humans at most economically valuable work.” What Yun’s proposal genuinely adds is that this is the first major Chinese lab publicly proposing an economic-output AGI definition on a Goldman conference stage — a positioning move for MiniMax’s international investor pitch, in a GDP-anchored variant that indexes to macro output rather than a fixed dollar profit target. The framing is a personal position at a conference, not a published MiniMax policy. Watch clause: whether other Chinese labs (Zhipu, Moonshot, DeepSeek) adopt the GDP-anchored framing publicly is the tell for whether this becomes a Chinese-lab discourse convention or stays a MiniMax-branded positioning line. Log against MOC - Major Companies.

  • 2026-08-26-AI-Digest — Short interest in MiniMax climbed to ~20% of free float ahead of earnings per S&P Global data cited by Bloomberg — the highest bearish positioning since MiniMax’s ~$620M January 2026 Hong Kong listing. Traders are pricing in margin compression from the Qwen / DeepSeek / Kimi K2 competitive envelope. Load-bearing framing the digest carries: do NOT frame this as a fresh Aug-2026 price war — there has been no new headline Qwen or DeepSeek API price cut in August; DeepSeek’s Aug-16 V4-flash / V4-pro repricing was a structural move to peak / off-peak tiers, not another round of nominal cuts. Margin compression is real but is a continuation of the H1 2026 dynamic, not a new event — the shorts are pricing continued grind, not a new inflection.

Key Developments

  1. A-Share IPO Filing on Top of Hong Kong Float (June 1, 2026): The single filing is straightforward; the pattern is the substance — MiniMax and Zhipu AI beat OpenAI and Anthropic to public markets in January, and the mainland listing layered on top of the Hong Kong float is the cleanest single-company instance of the Chinese frontier-lab capital-markets pattern. Drivers: US-listing barriers, abundant domestic capital, and the same unit-economics-validation logic pushing OpenAI‘s September IPO from the US side.

  2. MiniMax M3 Open-Weight Sparse-Attention Launch (June 2, 2026): The MiniMax Sparse Attention architecture with ~1/20th compute at 1M tokens (9× faster input, 15× faster generation vs dense), trained on 100T interleaved multimodal tokens, with weights set to drop to HF/GitHub within 10 days. Vendor-published benchmarks (SWE-Bench Pro 59%, BrowseComp 83.5) are unaudited but credible-as-claimed — the actual open-weights story of the week if even half of the sparse-attention efficiency holds at scale.

See also: MiniMax-M2, Vast, DeepSeek, MOC - Major Companies.