Local AI

Microsoft expands Phi language model family with new reasoning capabilities

Microsoft has introduced three new small language models (SLMs) focused on complex reasoning tasks: Phi-4-reasoning, Phi-4-reasoning-plus, and Phi-4-mini-reasoning. These models represent a significant advancement in what small AI models can accomplish, particularly in mathematical reasoning and multi-step problem solving. The flagship Phi-4-reasoning-plus, a 14-billion parameter model, demonstrates performance that rivals much larger AI systems. According … Read more

Alibaba launches Qwen3 models with competitive AI reasoning capabilities

Alibaba has released Qwen3, a new family of large language models that compete with leading AI systems from OpenAI and Google. The lineup includes two mixture-of-experts (MoE) models and six dense models, with parameters ranging from 0.6 billion to 235 billion. According to benchmarks shared by Alibaba, the flagship Qwen3-235B-A22B model outperforms DeepSeek R1 and … Read more

Pleias launches small reasoning models optimized for RAG with built-in citations

French AI startup Pleias has released two open-source small reasoning models specifically designed for retrieval-augmented generation (RAG) with native citation support. As reported by Carl Franzen for VentureBeat, the new models—Pleias-RAG-350M and Pleias-RAG-1B—are available under the Apache 2.0 license, allowing commercial use. Despite their small size, the models outperform many larger alternatives on multi-hop reasoning … Read more

Google’s Gemma 3 models now run on consumer GPUs through quantization

Google has released new versions of its Gemma 3 AI models that can run on consumer-grade graphics cards through a technique called Quantization-Aware Training (QAT). This development makes powerful AI models accessible to users without high-end hardware. The company announced that QAT dramatically reduces memory requirements while maintaining high quality performance. Gemma 3’s largest 27B … Read more

Nous Research launches AI model with optional reasoning mode

Nous Research has released DeepHermes-3, a new AI language model that allows users to switch between detailed reasoning and quick responses. As reported by Carl Franzen for VentureBeat, this 8-billion parameter model builds on Meta’s Llama technology. Users can activate a special reasoning mode that makes the AI show its thought process before providing answers. … Read more

Mistral Small 3 rivals larger competitors

French startup Mistral AI has announced the release of Mistral Small 3, a 24-billion-parameter language model that the company claims matches the performance of models three times its size. According to Mistral AI, the new model achieves 81% accuracy on standard benchmarks while processing 150 tokens per second, making it comparable to Meta’s Llama 3.3 … Read more

DeepSeek-R1 brings significant cost reduction for Enterprise AI

DeepSeek’s new AI reasoning model R1 could substantially reduce the costs of developing AI applications. According to an analysis by Ben Dickson in VentureBeat, DeepSeek-R1 offers similar capabilities to leading models at a fraction of the price. The model costs $2.19 per million output tokens, compared to OpenAI’s o1 at $60 per million tokens. When … Read more

Hugging Face launches compact AI models for image and text analysis

Hugging Face has released two new AI models designed for processing images, videos, and text on devices with limited resources. As Kyle Wiggers reports for TechCrunch, the models called SmolVLM-256M and SmolVLM-500M require less than 1GB of RAM to operate. The models, containing 256 million and 500 million parameters respectively, can describe images, analyze video … Read more

DeepSeek releases new reasoning models and introduces distilled versions

Chinese AI company DeepSeek has announced the release of its new reasoning-focused language models DeepSeek-R1-Zero and DeepSeek-R1, along with six smaller distilled versions. The main models, built on DeepSeek’s V3 architecture, feature 671 billion total parameters with 37 billion activated parameters and a context length of 128,000 tokens. According to company statements, DeepSeek-R1 achieves performance … Read more

Diffbot launches new AI model with real-time fact checking

Diffbot, a Silicon Valley company, has introduced a new AI model that combines AI capabilities with real-time fact verification. As reported by Michael Nuñez for VentureBeat, the system uses graph retrieval-augmented generation (GraphRAG) technology based on Meta’s Llama 3.3. The model connects to Diffbot’s Knowledge Graph, a database containing over one trillion facts that updates … Read more