AIME

AIME G500 - Multi GPU Workstation | AIME AIME G500 - Workstation Der AIME G500 ist als wartungsfreundlich High-End-GPU Workstation konzipiert, mit eine herausragende Kühlleistung und Netzteil-Kapazität, um bis zu vier High-End-GPUs zu betrei...

The AIME G500 is now even more powerful, supporting the Threadripper Pro 99xxWX CPUs!

www.aime.info/de/shop/prod...

Chinese company Z.ai released their model GLM-4.5 open source, a series models are foundation models designed for intelligent agents.

👉 GLM-4.5: 355B total / 32B active parameters

👉 GLM-4.5-Air: 106B total / 12B active parameters

github.com/zai-org/GLM-...

GitHub - MoonshotAI/Kimi-K2: Kimi K2 is the large language model series developed by Moonshot AI team Kimi K2 is the large language model series developed by Moonshot AI team - MoonshotAI/Kimi-K2

The chinese Company MoonshotAI released Kimi K2 as a state-of-the-art mixture-of-experts (MoE) language model with 32 billion activated parameters and 1 trillion total parameters.

github.com/MoonshotAI/K...

GitHub - wasserth/TotalSegmentator: Tool for robust segmentation of >100 important anatomical structures in CT and MR images Tool for robust segmentation of >100 important anatomical structures in CT and MR images - wasserth/TotalSegmentator

TotalSegmentator is a tool for segmentation of most major anatomical structures in any CT or MR image, created by the department of Research and Analysis at University Hospital Basel.

github.com/wasserth/Tot...

Black Forest Labs - Frontier AI Lab Amazing AI models from the Black Forest.

Black Forest Lab released FLUX.1 Kontext [dev], which delivers proprietary-level image editing performance in a 12B parameter model that can run on consumer hardware.

bfl.ai/announcement...

Reader API Convert any URL to Markdown for better grounding LLMs.

ReaderLM-v2 is a 1.5 B-parameter language model specialized for HTML-to-Markdown conversion and HTML-to-JSON extraction.

jina.ai/reader

Jina Embeddings v4: Universal Embeddings for Multimodal Multilingual Retrieval Jina Embeddings v4 is a 3.8 billion parameter universal embedding model for multimodal and multilingual retrieval that supports both single-vector and multi-vector embedding outputs.

Jina AI released Jina Embeddings v4, a 3.8 billion-parameter multimodal and multilingual embedding model based on the Qwen 2.5 VL 3B Instruct backbone. The architecture combines text and images in a unified semantic space.

jina.ai/news/jina-em...

GitHub - Tencent-Hunyuan/Hunyuan3D-2.1: From Images to High-Fidelity 3D Assets with Production-Ready PBR Material From Images to High-Fidelity 3D Assets with Production-Ready PBR Material - Tencent-Hunyuan/Hunyuan3D-2.1

Tencent Hunyuan3D-2.1 is a scalable 3D asset creation system that advances state-of-the-art 3D generation through two pivotal innovations: Fully Open-Source Framework and Physically-Based Rendering (PBR) Texture Synthesis.

github.com/Tencent-Huny...

GitHub - MiniMax-AI/MiniMax-M1: MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model. MiniMax-M1, the world's first open-weight, large-scale hybrid-attention reasoning model. - MiniMax-AI/MiniMax-M1

Minimax is making a bold move, releasing its large language model M1 that is even said to rival DeepSeek under an open source Apache License 2.0

github.com/MiniMax-AI/M...

🚀 AIME G500 Blackwell Pro AI Workstation 🚀

The AIME G500 is the most powerful HPC workstation worldwide, designed as maintainable high-end workstation with enough cooling and PSU capacity to host up to four high-end GPUs.

AI-powered coding for the enterprise | Mistral AI Lightning-fast completions, deep code understanding, and agentic software engineering—right where you work.

Mistral Code is an AI-powered coding assistant that bundles powerful models, an in-IDE assistant, local deployment options, & enterprise tooling into 1 fully supported package, so developers can 10x their productivity with the full backing of their IT and security teams.

mistral.ai/products/mis...

deepseek-r1 DeepSeek-R1 first-generation of open reasoning models with comparable performance to OpenAI-o3.

Deepseek released new model weights: In this update, DeepSeek R1 has significantly improved its reasoning and inference capabilities.

The new upgraded Deepseek R1 is now almost on par with OpenAI's O3 High model on LiveCodeBench.

ollama.com/library/deep...

Devstral | Mistral AI Introducing the best open-source model for coding agents.

Mistral released Devstral, an agentic LLM for software engineering tasks. Devstral is built under a collaboration between Mistral AI & All Hands AI. It outperforms all open-source models on SWE-Bench Verified by a large margin and was released under the Apache 2.0 license.

mistral.ai/news/devstral

Deep Learning GPU Benchmarks An overview of current high end GPUs and compute accelerators best for deep and machine learning and model inference tasks. Included are the latest offerings from NVIDIA: the Hopper and Blackwell GPU ...

We updated our GPU Benchmark blog article, adding RTX 5090 and RTX Pro 6000 Blackwell WS.

www.aime.info/blog/en/deep...

GitHub - QwenLM/WorldPM Contribute to QwenLM/WorldPM development by creating an account on GitHub.

Qwen WorldPM-72B (World Preference Modeling) demonstrates that preference modeling follows similar scaling laws as language modeling.

Through large-scale training on 15M preference data, they reveal that preference models can learn unified preference representations.

github.com/QwenLM/WorldPM

The 2025 AI Index Report | Stanford HAI

Stanford released The 2025 AI Index Report.

The AI Index offers one of the most comprehensive, data-driven views of artificial intelligence.

hai.stanford.edu/ai-index/202...

GitHub - xdit-project/xDiT: xDiT: A Scalable Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism xDiT: A Scalable Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism - xdit-project/xDiT

xDiT is a scalable inference Engine for diffusion transformers (DiTs). Think of it as vLLM for Pixels (IMages, Video).

xDiT is an inference engine designed for the parallel deployment of DiTs on a large scale.

github.com/xdit-project...

GitHub - ace-step/ACE-Step: ACE-Step: A Step Towards Music Generation Foundation Model ACE-Step: A Step Towards Music Generation Foundation Model - ace-step/ACE-Step

ACE-Step is an open-source foundation model for music AI that empowers artists, producers, and creators to generate music from text prompts.

github.com/ace-step/ACE...

GitHub - mcp-use/mcp-use: mcp-use is the easiest way to interact with mcp servers with custom agents mcp-use is the easiest way to interact with mcp servers with custom agents - mcp-use/mcp-use

MCP-Use helps to connect any LLM to any MCP server and to build custom agents that have tool access, without using closed source or application clients.

Let developers easily connect any LLM to tools like web browsing, file operations, and more. It is the open source.

github.com/mcp-use/mcp-...

GitHub - XiaomiMiMo/MiMo: MiMo: Unlocking the Reasoning Potential of Language Model – From Pretraining to Posttraining MiMo: Unlocking the Reasoning Potential of Language Model – From Pretraining to Posttraining - XiaomiMiMo/MiMo

Xioami released their open source LLM named MiMo.

MiMo-7B is a series of models trained from scratch for reasoning tasks, surpassing much larger 32B models.

github.com/XiaomiMiMo/M...

tngtech/DeepSeek-R1T-Chimera · Hugging Face We’re on a journey to advance and democratize artificial intelligence through open source and open science.

TNG Tech created a LLM-Chimera by merging DeepSeek-R1 and DeepSeek-V3, combining the reasoning capabilities of R1 with the token efficiency improvements of V3.

In benchmarks, it appears to be as smart as R1 but much faster, using 40% fewer output tokens.

huggingface.co/tngtech/Deep...

Bytedance (TikTok) released Vidi: Large Multimodal Models for Video Understanding and Editing that addresses video editing challenges by enabling temporal retrieval—identifying time ranges in hour-long videos linked to text queries.

bytedance.github.io/vidi-website/

GitHub - QwenLM/Qwen3: Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud. Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud. - QwenLM/Qwen3

Alibaba released Qwen3 LLM

Their new dense and Mixture-of-Experts (MoE) models (0.6B, 1.7B, 4B, 8B, 14B, 32B, 30B-A3B, 235B-A22B) enable seamless switching between reasoning-intensive tasks (math, coding) and efficient general-purpose chat.

github.com/QwenLM/Qwen3

S3MOT: Monocular 3D Object Tracking with Selective State Space Model

A new study introduces advancements in monocular 3D multi-object tracking (MOT) to address challenges in spatiotemporal association from 2D video streams.

Paper: lnkd.in/eXjpBhq7
Code: lnkd.in/eivtvnZh

GitHub - SakanaAI/AI-Scientist-v2: The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search - SakanaAI/AI-Scientist-v2

The AI Scientist-v2 is a generalized e2e agentic system that has generated the first workshop paper written entirely by AI and accepted through peer review.

This system autonomously generates hypotheses, runs experiments, analyzes data, & writes scientific manuscripts.

github.com/SakanaAI/AI-...

Gemma 3 QAT Models: Bringing state-of-the-Art AI to consumer GPUs- Google Developers Blog Explore Gemma 3 models now offering state-of-the-art AI performance on consumer GPUs with new int4 quantized versions optimized with Quantization Aware Training (QAT).

Google released Gemma 3 QAT Models, introducing Quantization-Aware Training (QAT), enabling state-of-the-art AI performance on consumer-grade GPUs.

developers.googleblog.com/en/gemma-3-q...

GitHub - microsoft/BitNet: Official inference framework for 1-bit LLMs Official inference framework for 1-bit LLMs. Contribute to microsoft/BitNet development by creating an account on GitHub.

Microsoft released a 1-Bit-DeepSeek derivate which runs on CPU.

github.com/microsoft/Bi...

GitHub - SkyworkAI/Skywork-OR1 Contribute to SkyworkAI/Skywork-OR1 development by creating an account on GitHub.

Skywork-OR1 (Open Reasoner 1) is a new SOTA 32B model family with open weights, training code and training data from China which includes two general-purpose reasoning models.

These models are fine tunings of deepseek-ai models.

github.com/SkyworkAI/Sk...

ChatGLM is a chinese family of LLMs from GLM-130B to GLM-4 All Tools.

The GLM-4-0414 Model Series were scaled up to 32B parameters, including models with capabilities for dialogue, reasoning, and rumination.

github.com/THUDM/GLM-4

Find their technical paper here: arxiv.org/pdf/2406.12793