ETH Zurich (Organization): news timeline & CVEs

One year later: Sovereign AI and the fight for choice

Cloudflare is bringing EuroLLM and Switzerland's fully open Apertus model to Workers AI.

Cloudflare said EuroLLM and Apertus are coming to Workers AI, with access available by request. EuroLLM, developed by a European university consortium with Horizon Europe and EuroHPC support, covers 35 languages including all 24 official EU languages and was trained on MareNostrum 5. Apertus, from ETH Zurich, EPFL and the Swiss National Supercomputing Centre, is fully open, trained on more than 15 trillion tokens and over 1,500 languages on the Alps supercomputer. Cloudflare also announced hands-on workshops to help government cyber agencies build model-agnostic AI defenses, starting in Singapore in October.

Cloudflare Blog13:04 UTC · 8h agoAI industry

AutoMark: Enabling Autoresearch to Discover Better LLM Watermarks

AutoMark uses frontier-model agents to autonomously discover distortion-free LLM watermarks that beat prior schemes.

AutoMark defines reliability criteria, statistical tests, and a suite ranking LLM watermarks by detectability, quality, and robustness so agents can search for new distortion-free schemes. Using GPT-6 Astra, Opus 5, and Gemini-3.8 Flash, the authors report more than 50 schemes, including several that outperform prior work on all three dimensions. A manual study decomposes the schemes and identifies new ideas, such as aligning watermark scores with a random per-request direction. Code is released at eth-sri/automark.

arXiv cs.CRSep 29, 11:45 UTC · 2d agoAI research

GGUF vs GPTQ vs AWQ vs EXL2: LLM Model Formats Explained (2026)

An explainer separates LLM storage containers (safetensors, GGUF) from quantization methods (GPTQ, AWQ, EXL2/EXL3), with memory math and quality trade-offs.

The article explains that containers like safetensors, GGUF, and PyTorch pickle define on-disk tensor storage, while GPTQ, AWQ, bitsandbytes NF4, and llama.cpp K/I-quants define quantization methods, with EXL2/EXL3 combining both. GGUF, introduced August 2023 by Georgi Gerganov, replaced GGML using typed key-value metadata and can embed tokenizers and chat templates; a Q4_K_M quant averages about 4.5 bits per weight. GPTQ, published at ICLR 2023 by Frantar, Ashkboos, Hoefler, and Alistarh, quantized a 175B-parameter model in roughly 4 GPU hours to 3-4 bits with ~3.25x speedup on A100 GPUs. It also notes safetensors eliminates pickle's arbitrary code execution risk when loading untrusted checkpoints.

MarkTechPostSep 19, 04:04 UTC · 12d agoAI tools & infra2

ETH Zurich

Timeline3 articles

Appears with

One year later: Sovereign AI and the fight for choice

AutoMark: Enabling Autoresearch to Discover Better LLM Watermarks

GGUF vs GPTQ vs AWQ vs EXL2: LLM Model Formats Explained (2026)