Neura News

AI News

News reporting focused on AI and machine learning, covering the companies behind these technologies, their real-world applications, and the ethical concerns they raise. This includes areas like generative AI (large language models, text-to-image and video), speech tech, and predictive analytics.

Latest News

14 articles

AI Models

DeepSeek Ships V4.1-Flash With 1M Context, MIT License, and a 763B Parameter Bill

DeepSeek released V4.1-Flash on September 10, 2026, an MIT-licensed open-weight model with a causal encoder-decoder architecture, native vision, and a 1M token context window. It carries 763B total parameters with 8B active for prefill and 16B for decode, and soft-retires V4 Pro at cheaper pricing. Artificial Analysis scored it 40 on its Intelligence Index at $0.27 per task, while Vals ranked it the top open-weight model.

Sep 1210 minNeura News

Product Launch

Acer Unveils Veriton RI110 Mini Workstation for Local Agentic AI

Acer unveiled the Veriton RI110 AI Mini Workstation on September 2, 2026, in Berlin. This compact desktop, featuring an Intel Core Ultra X7 processor and Intel Arc B390 graphics, supports local inference of AI models up to 120 billion parameters. It is designed for hybrid agentic AI workloads, combining local processing with cloud resources, and includes the Qubi Claw software suite for secure, autonomous AI tasks. The system offers up to 96 GB of LPDDR5X memory, 4 TB of SSD storage, and extensive connectivity options including OCuLink, Wi-Fi 7, and dual LAN ports. Availability begins in North America in Q4 2026 and EMEA in Q1 2027.

Sep 24 minNeura News

Research

Modular Pretraining: A New Approach to Containing Dangerous AI Knowledge

Researchers at Anthropic and AE Studio have introduced Gradient Routed Auxiliary Modules (GRAM), a method that isolates dangerous knowledge in large language models into switchable modules during training. This approach allows operators to control access to sensitive content, potentially reducing risks of misuse. Preliminary experiments show promise across models up to 5B parameters, but the method has not yet been applied to production-scale systems.

Aug 1712 minNeura News

AI Models

Nvidia's Nemotron 4 Targets Trillion-Parameter Scale, but Chinese Rivals Already Lead

Nvidia is developing Nemotron 4, an open-weight AI model family with a trillion-parameter flagship, expected as early as fall 2026. However, Chinese labs like Moonshot AI and DeepSeek already ship larger models, with Kimi K3 at 2.8 trillion parameters and DeepSeek V4 Pro at 1.6 trillion. Nvidia has tripled cloud spending to $28 billion through 2031, betting on open models to drive GPU demand, even as it competes with customers like OpenAI.

Aug 123 minNeura News

AI Models

ByteDance reportedly trains AI model with up to 10 trillion parameters

ByteDance, the parent company of TikTok, is reportedly training an AI model with up to 10 trillion parameters, which would make it the largest Chinese AI model ever built. The model, currently in pretraining, is expected to rival Anthropic's top system in scale. ByteDance has avoided distillation for over a year, focusing on original training data.

Aug 73 minNeura News

AI Models

Alibaba Releases Qwen3.8-Max, Claiming Frontier-Level Performance

Alibaba released Qwen3.8-Max, its largest and most capable AI model, claiming it rivals top US systems from Anthropic and OpenAI. The model, with 2.4 trillion parameters, ranks near the top on Arena.AI, and Alibaba will release its weights next week, marking a return to open-weight strategy amid intensifying US-China AI competition.

Aug 36 minNeura News

Research

Optical Receiver Edits Robot AI Memory Using Light Beams

Cornell Tech researchers have developed an optical receiver that directly alters memory using photocurrents from a projected light matrix. The technology could update AI model parameters in robots and data centers with lower energy consumption than traditional electrical links, though commercial use remains distant.

Jul 264 minNeura News

AI Models

Poolside Releases Laguna S 2.1, a Small Open-Weight Coding Model

Poolside has released Laguna S 2.1, its third coding model in three months. The mixture-of-experts model uses 8 billion active parameters and focuses on better behavior during long agentic sessions. It outperforms other agentic coding models in its weight class and sometimes approaches systems 10 to 20 times its size.

Jul 236 minNeura News

AI Models

Kimi K3 Open Model Nears GPT-5.6 Sol and Fable 5 Performance

Kimi has released K3, a multimodal open-weight model with 2.8 trillion parameters and a one million token context window. In benchmarks, it approaches Claude Fable 5 and GPT-5.6 Sol but trails them slightly. Priced at $3 per million input tokens and $15 per million output tokens, K3 signals an end to ultra-cheap Chinese AI models, though it remains cheaper than top Western competitors.

Jul 165 minNeura News

AI Models

Thinking Machines Releases 975B Parameter Inkling Model

Thinking Machines Lab, the startup founded by former OpenAI CTO Mira Murati, has released Inkling, an open-weights multimodal model with 975 billion parameters. The model leads U.S. open-source models on the Artificial Analysis Intelligence Index but trails top Chinese models. It shows strong agentic performance but has a 63 percent hallucination rate.

Jul 164 minNeura News

Product Launch

Xiaomi MiMo-V2.5-Pro Builds Compiler in 4.3 Hours

Xiaomi launched MiMo-V2.5-Pro, an open-weight mixture-of-experts model with 1.02 trillion parameters. It completed a full compiler project in 4.3 hours during internal tests and performs close to Claude Opus 4.6 on coding benchmarks. The model uses 40 to 60 percent fewer tokens than competitors like Claude Opus 4.6 and Gemini 3.1 Pro.

May 34 minNeura News

AI Models

IBM Granite 4.1 LLMs: Full Training Breakdown

IBM's Granite Team details the construction of Granite 4.1, a set of decoder-only LLMs in 3B, 8B, and 30B sizes trained on about 15 trillion tokens through five pre-training phases. The models receive supervised fine-tuning on 4.1 million curated samples and multi-stage reinforcement learning. The 8B instruct version equals or beats the prior Granite 4.0-H-Small 32B-A9B MoE model using a basic dense setup with fewer parameters. All models appear under Apache 2.0 license.

Apr 298 minNeura News