MATRIX · GLM models

A one-direction weight edit strips refusal from the shipped GLM-5.3-Flash weights

Fri, 11 Sep 2026 19:30:00 GMT

How Fragile Is Safety Alignment at Frontier Scale? (arXiv 2609.09793, Thu 10 Sep digest) extends directional ablation — removing a model's ability to refuse by projecting a single 'refusal direction' out of the weights, needing only a few hundred contrastive prompts and no optimization — from dense models up to ~70B to a frontier MoE release: GLM-5.3-Flash (320B parameters, 288 routed experts, four-wide hyper-connection residual, block-FP8 weights). The attack survives the architecture and the shipped quantization; with multiple experts and a hyper-connected residual stream, refusal-direction mass is distributed differently than in dense models, and the paper maps where refusal behavior actually lives in the shipped weights.

Recommendation: WATCH

Sources: arXiv 2609.09793

Z.ai open-sources GLM-5.3-Flash (new base, hybrid attention, MIT)

Fri, 28 Aug 2026 00:31:03 GMT

Z.ai released GLM-5.3-Flash weights under MIT on Hugging Face: a 320B-total / 18B-active natively multimodal MoE, the first GLM-5-series model trained from a newly built base rather than GLM-5.2. It combines sparse and linear attention to cut long-context serving cost and adds Manifold-Constrained Hyper-Connections, pre-trained on a 30T-token multimodal corpus. Z.ai claims it beats GLM-5.2 across benchmarks at one-tenth the price and approaches Claude Opus 4.8 on coding and agentic tasks; Artificial Analysis independently scored it 57 on their Intelligence Index (added 2026-08-26). vLLM, SGLang, Transformers, KTransformers and vLLM-Ascend support shipped day-0.

Recommendation: TRIAL

Sources: huggingface.co · GLM 5.3 Flash

Zhipu releases GLM-5.3, claiming the strongest open-weights coding model

Sat, 15 Aug 2026 12:26:05 GMT

Zhipu released GLM-5.3, built on the same base as GLM-5.2 with all gains coming from post-training ('scaling post-training is all we did'). It is positioned as the strongest open-weights coding model, with a heavy cybersecurity and agentic focus: working with Chinese security teams, it found 2,436 vulnerabilities across 269 projects, published in the public registry cvd.z.ai. The model is available now via the GLM Coding Plan, which works with Claude Code and OpenCode. The API is 'coming soon', and the open weights are planned for ~2 weeks out, after security reviews.

Recommendation: TRIAL

Sources: z.ai · glm 5.3