Z.ai Releases GLM-5.3-Flash as an Open-Weights Multimodal Model

LinkZ.aiTool

Z.ai has launched GLM-5.3-Flash under the MIT licence on Hugging Face, marking the first natively multimodal model in the GLM-5 family. Featuring 320B total parameters and 18B active parameters, it targets high efficiency across coding and agentic tasks.

Key points

  1. Total size of 320 billion parameters with only 18 billion active parameters during inference.
  2. Hybrid architecture pairing sparse and linear attention to lower serving costs for long contexts, alongside Manifold-Constrained Hyper-Connections (mHC).
  3. Pre-trained on a multimodal corpus consisting of 30 trillion tokens.
  4. Native deployment support across frameworks including vLLM, SGLang, Transformers, Unsloth, TokenSpeed, and KTransformers.
  5. Configurable reasoning budget via the reasoning_effort parameter (low, high, max; defaults to max) and clear_thinking toggle for chat interactions.
  6. Published under the MIT licence with weights in Safetensors format (FP8, BF16, F32).

Why it matters

Hybrid-attention and sparse activation architectures significantly decrease serving overhead for long-context multimodal models on self-hosted infrastructure. The permissive MIT licence removes legal and commercial barriers for teams deploying local or private agents.

Analysis

Z.ai positions GLM-5.3-Flash as a cost-efficient model capable of approaching Claude Opus 4.8 in coding and agentic benchmarks while outperforming GLM-5.2. The integration of linear attention and hyper-connections addresses practical serving bottlenecks, and explicit parameters for reasoning budget management give developers direct control over token generation costs.

Mentioned links

Found in

Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • Article:docs.z.ai

    Z.AI Launches GLM-5.3-Flash Featuring Native Multimodality

    Z.AI has introduced GLM-5.3-Flash, the first native multimodal model in the GLM-5 series. Featuring 320 billion total parameters with 18 billion activated, it leverages a hybrid sparse and linear attention architecture to cut compute and memory requirements.

    AI & LLMs· Announcement

  • Link:Z.ai

    Z.ai Releases GLM-5.2 with 1M Context Window and MIT Licence

    Z.ai has released GLM-5.2 on Hugging Face under the permissive MIT licence. The flagship open-weights model is designed for long-horizon agentic and coding workflows, featuring a robust 1M-token context window. Architectural updates including IndexShare and an enhanced MTP layer significantly cut computation FLOPs and accelerate speculative decoding.

    AI & LLMs· Resource

  • Link:Moonshot AI

    Moonshot AI Releases Kimi K3: Open-Weight 2.8T MoE Model

    Moonshot AI has released Kimi K3, an open-weight native multimodal agentic model featuring 2.8 trillion total parameters and a 1-million-token context window. The architecture uses a sparse MoE setup that activates 104 billion parameters per token.

    AI & LLMs· Announcement

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.