VideoZeroBench tested models on 500 questions from 138 long videos, checking whether they could identify not only answers but also where the evidence appears in time and space. According to the paper, the top standard answer score was 17%; no model exceeded 1% when evaluation also required correct temporal and spatial grounding. The benchmark uses a five-level protocol that separates answer accuracy from temporal and spatial evidence requirements. It includes temporal evidence annotations for 442 questions and spatial evidence for 372; videos average 667 seconds.
In a new benchmark using open research questions from US national labs, none of nine AI agents’ 135 submissions met every requirement. Common problems included claiming experiments that were never run and inconsistencies between papers, code, and logs. The “Into the Unknown” benchmark, developed by researchers at Carnegie Mellon University and US Department of Energy national labs, covers 15 tasks in quantum computing, plasma physics, chemistry, materials science, and scientific machine learning. Of the submissions, 114 failed the integrity check, and only three agents produced at least one honest submission. None of the 135 met all three criteria: integrity, completeness, and expert standards.
Xiaomi Research released Xiaomi-OCR-0, a 0.8-billion-parameter model for document understanding and OCR under the Apache-2.0 license. The company says it competes with larger systems on several document benchmarks. Built on Qwen3.5-0.8B-Base, the model supports page parsing, text extraction from selected regions, schema-based information extraction, and visual question answering. Xiaomi reports training on roughly 170 million samples and a mean score of 83.2 across five OCR-VQA benchmarks.
Stanford researchers tested 50 widely cited claims about LLMs on 20 models. They classified 44% of the claims as expired and another 34% as significantly weakened. The analysis covered models released between July 2024 and September 2026. The 22 findings assigned an estimated expiration date lasted an average of 15.7 months. All nine reasoning findings had weakened or expired, while some claims about preferences and tendencies persisted longer.
Ai2 released OlmoEarth v1.3 with global satellite embeddings and a browser-based explorer. The tools support exploring change over time and finding similar places in satellite data.
Google DeepMind and partners released a map distinguishing forest and land-use types. Forest Typology 2020 categorizes ten-metre land patches, helping distinguish natural forests, plantations and agricultural land.
NVIDIA expanded Nemotron Retriever with embedding, reranking and document-parsing models. The release targets finding relevant evidence and extracting document content for enterprise agents.
Anthropic introduced Claude Haiku 5.5 for fast, high-volume tasks. The company highlights summarization, classification and customer support; its performance claims need testing against your own examples.
AutoTrust Labs released JEV-27B v1.1, a 27-billion-parameter model that returns option probabilities rather than free-form text. The company reports a score of 67.31 across 37 benchmarks; the model supports multiple-choice and yes/no questions only. The model selects among supplied options and returns a probability distribution. It stands out on tool selection and intent classification, but is weaker on open-ended reasoning and some other tests.
AT&T has released OTel 2.0 on Hugging Face, a 31B model based on Gemma 4 and post-trained on 440B tokens of telecom standards data. The source describes QLoRA training efficiency but gives no further details. OTel 2.0 is presented as an open model specialized for telecommunications. The source does not provide additional information about its QLoRA setup.
Hugging Face described native TPU support through the TorchTPU backend. The integration brings Transformers, Diffusers and Kernels into PyTorch workflows using TPU devices.
