NVIDIA published a tutorial showing how to serve Z.ai's GLM-5.3 NVFP4 model, a 753-billion-parameter mixture-of-experts model with 40 billion parameters activated per token, across two GB300 DGX Stations linked via ConnectX-8 RDMA. Because the roughly 464 GB quantized checkpoint exceeds the 252 GB of HBM3e on a single GB300 GPU, the setup uses tensor parallelism (TP=2) and vLLM to split weights and computation across both machines while exposing one OpenAI-compatible API endpoint, reportedly measuring throughput of 607.8 tokens per second.
developer.nvidia.com
· 2026-11-14
Tom's Hardware ran deeper AI benchmarks comparing Apple's Mac Studio with M5 Ultra against Nvidia's GB10-based DGX Spark system, this time loading large models such as DeepSeek V4.1 Flash and GLM 5.3 Flash to make use of the Mac's 256GB-plus memory pool. The tests included Artificial Analysis' AgentPerf-Local benchmark, which replays agentic AI workloads weighted heavily toward prompt processing rather than token generation.
tomshardware.com
· 2026-11-10
Researchers report that GLM 5.3, an open-weight AI model released by Z.ai in August, was able to identify and exploit Firefox vulnerabilities during the gap between a security fix being published and reaching users. They found that Firefox versions released between September 17 and October 26 could be compromised this way, with the model building a malicious webpage that silently stole a file from a visitor's computer, at a cost of roughly $1,000 per version. The researchers disclosed the issue to Mozilla, which shipped Firefox 159.0 on October 27, 2026 patching a previously undisclosed vulnerability the model had also used.
epoch.ai
· 2026-10-29
Mistral has published the full weights for Mistral Large 4, the trillion-parameter model nicknamed 'Le Chonk' that first appeared in preview on Oct. 6. New benchmark results show its score on the Terminal-Bench 4.0 coding test rising from 28% to 41%, putting it slightly ahead of GLM-5.3 on that metric, though GLM-5.3 and Codex GPT-6.1 Terra still edge it out on Artificial Analysis's overall composite index. Mistral also says the model uses 47% fewer tokens per task on average than comparable open models.
zdnet.com
· 2026-10-29
Mistral has released the downloadable weights for Mistral Large 4, a trillion-parameter open-weight AI model nicknamed 'Le Chonk,' letting businesses and governments deploy the system on their own infrastructure. The company says it matches the performance of China's GLM-5.3 while using 47% fewer tokens per task, and claims it leads all open models on Artificial Analysis' Coding Agent Index.
techcrunch.com
· 2026-10-29
Cloudflare disclosed that its GLM-5.3 model on Workers AI exposed cached data from other customers' requests to 155 responses between August 27 and September 9, 2026, affecting only prompts longer than 681,280 tokens. A security researcher from Knostic reported the bug on September 8; Cloudflare took the model offline within 40 minutes of confirming the incident and restored it the next day after patching its SGLang inference engine.
blog.cloudflare.com
· 2026-10-26
Neuralwatt CEO Chad Gibson analyzed a month of cloud traffic and found that different agent harnesses produce wildly different prompt-cache hit rates, from 1.6% to 11.1% uncached input tokens. The study found that a cache miss roughly doubles a request's energy use, and that deep sessions past 128k tokens with major prompt re-reads can cost as much compute as roughly 600 normal turns at that depth.
neuralwatt.com
· 2026-10-16
Ultramarine 1 Flash is a new model priced at $0.13 per task on internal benchmark PI Knowledge Work, which the company says correlates strongly with the Artificial Analysis Intelligence Index. It is built on GLM-5.3 Flash and generates 6.5% fewer tokens than that base model while outscoring Claude Fable 5 (max) and GPT-5.6 Sol (max), which cost $8.19 and $2.09 per task respectively. Pricing on the Ultramarine API is $0.15 per million input tokens, $0.03 for cached input, and $0.50 for output.
paperinstruments.com
· 2026-10-15
A new open-source benchmark called the Telegraph Test shows that instructing large language models to answer in 'cablese' — the clipped, article-free style once used by telegraph operators — reduces billed output tokens by 40-49% on several models' own API meters, while preserving or even slightly improving factual recall. The effect held across four model families tested with roughly 1,300 questions over 50 passages, with one exception: gpt-5-mini's reasoning overhead made cablese prompting roughly double its cost instead of cutting it.
fiveminutesforward.com
· 2026-10-07
A development team set out to run an entire month of coding work on the GLM 5.3 Flash model, keeping within a $68 budget for the first two weeks. The second half of the month saw usage spike to 1 billion tokens across other models, driven by an expensive vibe-coded Wagtail MCP server prototype that alone cost $150 and 5kWh, plus infrastructure capacity issues that forced switches to DeepSeek V4.1 Flash and Qwen 3.8 Flash.
wagtail.org
· 2026-10-02
Anthropic published benchmark results claiming Zhipu AI's open-weight GLM-5.3 model can autonomously generate working cyberattack exploits, including chained exploits, with safeguards that can reportedly be bypassed. In its Exploitbench tests, GLM-5.3 produced 50 successful Chrome exploits out of 410 attempts, close to the 56 achieved by Anthropic's own unreleased Claude Mythos model, while rivals Kimi K3 and DeepSeek V4.1 Flash scored 0% on a related control-flow hijack test.
tomshardware.com
· 2026-09-30
Anthropic reports that GLM-5.3, a new model from China's Zhipu AI (Z.ai), can autonomously build end-to-end cyber exploits similar to Anthropic's own Claude Mythos Preview. Anthropic says simple jailbreak techniques bypassed GLM-5.3's safeguards between 64% and 100% of the time in simulated tests, whereas the same attacks failed against safeguarded Claude models. NIST's Center for AI Standards and Innovation separately assessed GLM-5.3 as the most cyber-capable open-weight model released so far, trailing the US frontier by about four months.
anthropic.com
· 2026-09-29