Chinese LLMs H1 2024 Sprint: Qwen 3.8, K3, GLM 5.2, Hy3, M3, DeepSeek‑V4, MiMo‑V2.5
The first half of 2024 saw Chinese large‑language‑model providers launch a rapid series of trillion‑parameter and multimodal models—including Kimi K3, Qwen 3.8, DeepSeek‑V4 and others—while climbing to dominate most spots in global leaderboards, yet they now face deeper engineering and commercial challenges.
July’s roundup highlights how Chinese large‑language‑model (LLM) vendors accelerated their releases in the first half of 2024, with Kimi K3 debuting the first open‑source model exceeding three trillion parameters, Alibaba’s Qwen 3.8 preview arriving, and DeepSeek V4 moving to official release.
Release timeline (2024 H1) :
January – Wenxin 5.0 official, Kimi K2.5.
February – Zhipu GLM‑5, ByteDance Seedance 2.0, MiniMax M2.5, Step 3.5 Flash, Qwen 3.5.
March – MiniMax‑M2.7, Qwen 3.6, Xiaomi MiMo‑V2‑Pro and MiMo‑V2‑Omni.
April – DeepSeek V4 preview, Tencent Hy3 preview, GLM‑5.1, Qwen 3.6.
May – Step 3.7 Flash, Xiaomi MiMo‑V2.5, Alibaba Qwen 3.7.
June – Zhipu MiniMax M3, GLM‑5.2, Kimi K2.7.
July – Tencent Hy3 official, Meituan LongCat‑2.0, Huawei openPangu‑2.0‑Flash open‑source, Kimi K3, Alibaba Qwen 3.8 preview, DeepSeek V4 official.
Beyond pure LLMs, multimodal, speech and embedding models also saw notable releases during this period.
Leaderboard analysis shows Chinese vendors occupying seven of the top‑10 slots (Xiaomi, DeepSeek ×2, Tencent ×2, Zhipu, MiniMax), while the United States is represented only by Anthropic and Nvidia. Kimi K3 scored 57 points, ranking third overall and being the sole Chinese model in the top‑10.
Although the sheer scale of parameters has exploded—now reaching the trillion‑level—the article stresses that scaling is merely the first step. Real‑world deployment hurdles are deeper: enterprises need models that integrate reliably with business systems, remain stable under load, and can handle accounting‑level calculations, not just solve benchmark tasks.
Commercially, the market is testing pricing strategies; API pricing has shifted from aggressive price wars to Kimi K3’s proactive price increase, reflecting an industry still searching for a sustainable “why charge” model. Customer willingness to pay, scenario fit, and ROI validation remain early‑stage concerns.
The author concludes that the next phase is less about catching up technically and more about which player can first overcome the combined barriers of technology, productization, and cash flow, as the “second half” elimination round is only beginning.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
AI Engineer Programming
In the AI era, defining problems is often more important than solving them; here we explore AI's contradictions, boundaries, and possibilities.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.