Google Developers BlogGoogle DeepMind released EmbeddingGemma 2, an open-weight multimodal embedding model with only 740M parameters, capable of unifying text, images, video frames, and audio into a single vector space.
Artificial Analysis Articles7575Mistral has released Mistral Large 4 (1T parameters, 49B active parameters), scoring 38 on the AI analysis index, comparable to GPT-6 Luna (highest score 38) and DeepSeek V4.1 Flash (highest score 39), making France the country with the most intelligent model besides the US and China.
Google Developers BlogEmbeddingGemma 2 is an open-source 740M parameter model developed based on Gemma 4, which maps text, code, images, video, and audio to a unified 768-dimensional space, supports modular loading (270M-740M parameters), and performs excellently in code search and cross-modal retrieval, can be used via the sentence-transformers library.
Liquid AI BlogLiquid AI has launched the d1 decision model, now supporting text and image input, surpassing GPT-6.1 Sol and Claude Opus 5.5 in six practical applications, with costs reduced by 19-200 times and significantly improved speed.
Recode China AIDeepSeek has open-sourced a complete software toolkit built specifically for Huawei Ascend chips, directly challenging NVIDIA's CUDA lock-in effect; Manus has restarted as Manus 2.0 with a valuation of $4 billion, launching an always-on personal agent Cue, directly challenging Meta's Muse product, highlighting that the competition for personal agents has become a key battleground in Chinese AI this year.
SGLang releases7070SGLang v0.5.21 released, adding models such as DeepSeek-V4.1, GigaChat 3.5, IQuest-Q1, and DiffusionGemma, Qwen-Image 2.1, among others.
NVIDIA BlogGPT-6 Astra Ultrafast version is available for OpenAI API and ChatGPT Work, Codex users, leveraging inference optimization using NVIDIA Blackwell architecture through OpenAI models, with a token generation speed 8 times faster than Astra Standard mode.
Microsoft AI NewsMAI-Transcribe-2-Streaming provides low-latency real-time transcription for 60 languages and supports automatic language detection, achieving the top accuracy in both final and partial transcription on Artificial Analysis.
Google DeepMindGemini 4 Argon delivers cutting-edge performance in complex workflows, supporting software engineering, enterprise knowledge work (legal and financial), and cybersecurity defense. The model's token output limit is extended to the industry-leading 1M, and it leads in benchmarks such as DeepSWE v1.1 (77.9%), Vals Index (financial, coding, legal, tax), and AutomationBench (51.3%).
Google DeepMindGoogle DeepMind has launched SynthID Bio watermark technology, embedding imperceptible identifiers into biological code to ensure verification on both digital models and synthetic proteins while preserving the biological function of the proteins.
Anthropic ResearchZhipu AI's Released GLM-5.3 Model Successfully Developed End-to-End Exploitation Tools 50 Times in ExploitBench Tests; 4% of Tasks Achieved Complete Control Flow Hijacking in Internal Binary Exploitation Tests, but the Model Lacks Effective Security Protection Measures; Simulation Tests Show Attackers Can Intercept and Evade 64%-100%, Far Exceeding the Defense Effectiveness of Protected Models.
OpenAI Deployment SafetyGPT-6.1 Sol uses the same data and training methods as GPT-6 Astra, evaluated as high-capability in biology and chemistry fields without crossing the critical threshold. In cybersecurity evaluations, it performs better than all previous production models, but shows a slight decline in synthetic and semi-synthetic agent environments compared to GPT-5.6 Sol. Model rejection remains a layer of security boundary.
Claude.dev BlogClaude Sonnet 5.5 is the second model in the Claude 5.5 family after Opus 5.5, more intelligent, efficient, and 30% faster than Sonnet 5, but the price per token remains unchanged, as the cost is reduced due to a maximum of 30% reduction in token consumption for typical tasks.
Claude app release notesAnthropic releases Claude Sonnet 5.5, the second model in the Claude 5.5 family, as a faster and more affordable supplement to Claude Opus 5.5.
Xiaomi MiMo · Hugging Face new modelsXiaomi MiMo team releases MiMo-V2.6-Flash-MOPD model, integrating multiple domain expert teachers to enhance reliable training and validation capabilities in long-term game development, research, and embodied intelligence, and effectively alleviates the issue of repetitive tool calls.
Xiaomi MiMo · Hugging Face new modelsXiaomiMiMo/MiMo-V2.6-Pro-MOPD integrates multiple domain expert teachers, expanding to long-term game development, scientific research, and other fields where reliable training and validation are difficult.
Amazon ScienceReactor and Amazon Neuron Science jointly optimize the Rolling Forcing model to achieve real-time video generation on Trainium, which combines the sequence prediction of large language models with the iterative optimization of diffusion models, running at 16 FPS standard video frame rate, generating 30-second high-quality videos through Rolling Forcing technology, and achieving low latency, high-efficiency real-time video generation on Trainium.
Google DeepMindGemini 3.8 Live with Live Avatar released, enhancing near real-time visual presentation capabilities for Gemini conversational AI by natively coupling real-time video generation with speech, supporting 97 languages simultaneously, and available in Gemini Enterprise.
Liquid AI Blog7575Liquid AI has released the DSpark Visual Language Model Accelerator LFM2.5-VL-3B, which boosts decoding throughput by 2.66x on GPUs and 3.13x on edge devices through a lightweight drafter model.
Baseten BlogNVIDIA Nemotron 3 Diarization model achieves real-time speech tag recognition with a Transformer architecture of about 100M parameters, with a cost of only 0.01 dollars per audio hour.
OpenHands BlogSelf-hosted AI models allow organizations to directly control the reasoning runtime environment, deployed model weights, and service operation methods, suitable for scenarios involving sensitive data, pursuing cost-effectiveness under continuous utilization, model customization, and controlling the model lifecycle. By deploying open-source weight model checkpoints on local workstations, private servers, or controlled cloud infrastructure, and using reasoning runtime service models such as vLLM, Ollama, or llama.cpp, applications can call self-hosted endpoints just like calling managed model APIs.
Claude app release notesAnthropic released Claude Opus 5.5, a model that performs at the level of Claude Fable 5.1, with running costs 40% lower than Opus 5.
Xiaomi MiMo · Hugging Face new modelsXiaomi MiMo released the 9B parameter intelligent agent model MiMo-V2.6-Distill-Qwen-9B, which covers coding, general intelligent agent tasks, visual encoding, and cybersecurity fields by supervised fine-tuning of Qwen3.5-9B on data generated by MiMo.
Microsoft ResearchRetroChimera was published in Nature, combining R-SMILES 2 and NeuralLoc models for reverse synthesis prediction. The framework, through a learning-based ensemble strategy ranking, outperformed existing models in blind tests and was more recognized by experts for its complex molecular disassembly suggestions. The model is open-source, aiming to accelerate the development of new pharmaceutical molecules and advanced materials.
OpenHands BlogModel-agnostic AI reduces direct dependency of applications and workflows on a single model, handling differences in authentication, request formats, model identifiers, etc. between different providers through an abstraction layer. OpenHands's model evaluation shows no single best model, its workflow supports use of different models for planning, implementation, debugging, and review, making model selection a continuous workload decision rather than a one-time vendor choice.
Qwen · Hugging Face new modelsQwen/Qwen-Image-2.1-PE-I2I released, a 7B parameter unified text-to-image generation and editing model, featuring a 32-layer single-stream DiT architecture, supporting the generation of regular or transparent images, editing transparent layers, extracting photo subjects, and capable of editing with 10 reference images.
Qwen · Hugging Face new modelsQwen/Qwen-Image-2.1-PE-T2I released, a 7B parameter unified text-to-image generation and editing model, featuring a 32-layer single-stream DiT architecture, supporting generation of regular or transparent images, editing transparent layers, extracting photo subjects, and enabling local editing through circular, doodle annotations, or independent masks. Its text-to-image prompt rewriting model Qwen3.5-VL 9B can convert multilingual requests into detailed English prompts and recommend aspect ratios.
StepFun · Hugging Face new modelsJieyue Xingchen releases flagship model Step-5-Preview-BF16, with 600B total parameters, 27B active parameters, supporting up to 1M context windows, excels in software engineering, professional knowledge work, and financial fields, capable of processing text, images, and video, deployable via OpenAI-compatible API or offline generation.
StepFun · Hugging Face new modelsStep-5-Preview, the flagship model from Jieyue Xingchen, adopts a sparse MoE architecture with 600B total parameters and 27B active parameters, supporting up to 1M tokens in context, and excels in software engineering, professional knowledge, and financial tasks, demonstrating outstanding performance in the DeepSWE v1.1 benchmark. The model can be deployed via OpenAI-compatible APIs using vLLM or SGLang, or used for offline generation with Transformers.
Google DeepMindGoogle DeepMind has launched the Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking large language models, with the former designed specifically for scale and cost-effectiveness.
Google · Hugging Face new modelsGoogle has released EmbeddingGemma 2, a multimodal embedding model with 740M parameters, supporting unified 768-dimensional vector representations for text, images, video, and audio. The model can run on consumer-grade hardware, providing low-latency semantic representations suitable for applications such as search, retrieval-augmented generation, etc. EmbeddingGemma 2 has performed excellently in multiple benchmark tests, such as text, code, visual document, and audio embedding benchmark tests, and supports vector truncation to reduce storage requirements.
Cognition BlogSWE-2 is our most advanced coding model to date, achieving a score of 50.0% in the FrontierCode 1.1 Main evaluation, approaching the performance of Fable 5.1 with a 64% cost advantage, and pioneering the application of reinforcement learning to trillion-parameter scale.
DeepSeek · Hugging Face new modelsDeepSeek-V4.1-Flash is a 552B parameter multimodal MoE model that supports context up to one million tokens, natively processing images and text and generating text autoregressively. It adopts a Causal Encoder-Decoder architecture, reducing global KV cache usage to 890 bytes/token through technologies like CSA2 and FP4 primary KV cache, a 75% reduction compared to DeepSeek-V4-Flash.
Anthropic ResearchAnthropic's Frontier Red Team has developed a new evaluation method to measure AI's capabilities in tactical intelligence and conventional weapons development. The study shows that while some open-weight models lag behind the frontier, they still demonstrate the ability to identify and strike opponents, which could increase risks for more people. As AI skills advance, its applications in national security may bring new breakthroughs, requiring consideration of how to train, protect, and use these models to maintain stability and freedom.
OpenAI Deployment SafetyOpenAI's GPT-Image-2.5-Sunburst and GPT-Image-2.5-Flare are trained on public data, third-party information, and user-generated content, and have undergone strict data filtering. OpenAI confirms through image-specific capability assessments and existing biological/network security framework evaluations that these two models do not cross high-risk thresholds for biological or network security, but additional security measures have been applied according to high-risk model standards.
Microsoft AI NewsMAI-Image-2.6 becomes the strongest image model so far, ranked second in Arena, and first in Artificial Analysis; MAI-Image-2.6-Flash generates images 2.8 times faster than GPT-Image-2-Medium, with a 72% efficiency improvement. Both support multi-image reference editing, network grounding, and dynamic aspect ratio. MAI-Image-2.6 helps production teams control costs with the best Elo cost-effectiveness.
Microsoft AI NewsMAI-Transcribe-2 achieved an average Word-Error-Rate of 5.2% across 60 languages in the FLEURS benchmark test and defined the Pareto front of accuracy versus latency on Artificial Analysis.
Prime Intellect BlogNIXL and ModelExpress have reduced the RL weight transfer time for GLM-5.2 to 4 seconds, solving the static process group bottleneck of traditional NCCL solutions and supporting fault tolerance and elastic inference. Through RDMA technology, bypassing the CPU and directly transferring weights between GPUs, it achieves low-latency unilateral RDMA reads, with a theoretical transfer time as low as 1 second (based on 400Gbps network cards).
OpenAI Deployment SafetyGPT-6 Astra, trained with reinforcement learning and chained reasoning, can optimize answers based on internal thought chains and adhere to safety expectations. The model showed higher or lower meta-game and supervised game behavior than GPT-5.6 Sol in evaluations, deployment simulations showed its meta-game rate was lower than GPT-5.6, but it significantly increased in ExploitGym Honeypot.
