🔄 [GitHub Releases] turboderp-org/exllamav3 - 0.0.42


https://github.com/turboderp-org/exllamav3/releases/tag/v0.0.42

🔄 [GitHub Releases] vllm-project/vllm - v0.23.0


https://github.com/vllm-project/vllm/releases/tag/v0.23.0

🔄 [GitHub Releases] sgl-project/sglang - v0.5.13


https://github.com/sgl-project/sglang/releases/tag/v0.5.13

📰 Anthropic - Statement on the US government directive to suspend access to Fable 5 and Mythos 5


https://www.anthropic.com/news/fable-mythos-access

🆕 [HF Models] inclusionAI - VISTA-9B

https://huggingface.co/inclusionAI/VISTA-9B


🆕 [HF Models] inclusionAI - VISTA-4B

https://huggingface.co/inclusionAI/VISTA-4B

🗓️ Weekly GitHub Activity


🦙 llama.cpp
└ Release: b9544 → b9627
└ 83 commits

- Support for EAGLE3 speculative decoding #18039
- Added Gemma 4 Multi-Token Prediction (MTP) and assistant draft-model support #23398 #24282
- New architecture support for Cohere2-MoE #24260
- Multi-Token Multi-Domain (MTMD) models now support video input and a batching API #24269 #24384
- WebUI implemented as a Progressive Web App (PWA) with offline caching #23871
- WebUI added an opt-in sandboxed JavaScript execution tool #24244
- GGML core bumped to version 0.15.1 e08c226
- WebGPU performance improvements for prefill and k-quants #24225
- Vulkan added fast paths for contiguous transfers and dot2 product extension support #23973 #24123
- Fixed CUDA ssm_scan_f32 data-races and CPU rms_norm_back in-place aliasing #24360 #24305
- Server added prompt logging to local directories #22031

🔗 All changes | Latest release


🎨 stable-diffusion.cpp
└ Release: master-679-f3fd359 → master-694-276025e
└ 15 commits

- Added circular RoPE support for ideogram4 #1627
- Introduced free_sd_images function to manage memory for C API #1633
- Optimized performance by capping planner budget when models exceed streaming limits #1612
- Normalized APG diff_norm calculations by tensor size #1620
- Fixed SD3 conditioning crash when clip_l text encoder is missing #1638
- Corrected mask shape for masked flash attention #1625
- Resolved LoKR application issue by correctly marking w2_a tensors #1650

🔗 All changes | Latest release


🤗 Fresh models trending on HuggingFace:

bosonai/higgs-audio-v3-tts-4b ♡414
nex-agi/Nex-N2-mini ♡193
prefeitura-rio/Rio-3.5-Open-397B ♡108
RazzzHF/Realism_Engine_Ideogram_4 ♡90
silx-ai/Quasar-Preview ♡63
mindlab-research/Macaron-V1-Preview-749B ♡57
Zyphra/ZONOS2 ♡56
BennyDaBall/Z-Image-Engineer-V6 ♡46
PaddlePaddle/pp-ocrv6 ♡43
MooreThreads/MusaCoder-27B ♡35
apodex/Apodex-1.0-mini ♡31
Muhammadreza/alduin-4b-it-base ♡28
zjunlp/LabVLA ♡26
fancyfeast/bigasp-3 ♡20
Photoroom/prxpixel-t2i ♡20
LatentForce-ai/Cassini-1.0 ♡20
Gryphe/Pantheon-Reasoning-26B-A4B-1.1 ♡19
libertywing/FlashMemory-Deepseek-V4 ♡19
dx8152/Flux2-Klein-9B-Migration ♡19
apodex/Apodex-1.0-4B-SFT ♡18
Gryphe/Gemma-4-31B-StyleTune ♡17
VAGOsolutions/SauerkrautLM-LFM2.5-GLiNER ♡16
tsolful/zjourney-Ideogram-4-Fantasy-Realism-Refiner ♡14

🔄 [GitHub Releases] turboderp-org/exllamav3 - 0.0.43


https://github.com/turboderp-org/exllamav3/releases/tag/v0.0.43

📰 OpenAI - New OpenAI Academy courses for the next era of work

https://openai.com/index/academy-courses-applying-ai-at-work


🔓 OpenAI - How an astrophysicist uses Codex to help simulate black holes

https://openai.com/index/using-codex-to-simulate-black-holes


🔓 OpenAI - BBVA puts AI at the core of banking with OpenAI

https://openai.com/index/bbva


🔓 OpenAI - Creating new simulations of black holes with Codex

https://openai.com/index/creating-new-simulations-black-holes


🔓 OpenAI - Ad Tools Terms

https://openai.com/policies/ad-tools-terms


🔓 OpenAI - OpenAI Academy

https://openai.com/academy


🔓 OpenAI - How Preply combines AI and human tutors to personalize learning

https://openai.com/index/preply

🆕 [HF Models] tencent - Hy-Embodied-0.5-VLA-RoboTwin

https://huggingface.co/tencent/Hy-Embodied-0.5-VLA-RoboTwin


🆕 [HF Models] tencent - Hy-Embodied-0.5-VLA-UMI

https://huggingface.co/tencent/Hy-Embodied-0.5-VLA-UMI

🆕 [HF Models] microsoft - FastContext-1.0-4B-SFT


https://huggingface.co/microsoft/FastContext-1.0-4B-SFT

🆕 [HF Models] swiss-ai - Apertus-v1.1-0.5B


https://huggingface.co/swiss-ai/Apertus-v1.1-0.5B

🆕 [HF Models] microsoft - FastContext-1.0-4B-RL


https://huggingface.co/microsoft/FastContext-1.0-4B-RL

🆕 [HF Models] swiss-ai - Apertus-v1.1-1.5B-Instruct

https://huggingface.co/swiss-ai/Apertus-v1.1-1.5B-Instruct


🆕 [HF Models] swiss-ai - Apertus-v1.1-4B-Instruct

https://huggingface.co/swiss-ai/Apertus-v1.1-4B-Instruct


🆕 [HF Models] swiss-ai - Apertus-v1.1-0.5B-Instruct

https://huggingface.co/swiss-ai/Apertus-v1.1-0.5B-Instruct


🆕 [HF Models] swiss-ai - Apertus-v1.1-4B

https://huggingface.co/swiss-ai/Apertus-v1.1-4B


🔓 [HF Models] swiss-ai - Apertus-v1.1-1.5B

https://huggingface.co/swiss-ai/Apertus-v1.1-1.5B

📰 LMSys - The next generation of speculative decoding: DFlash and Spec V2


https://lmsys.org/blog/2026-06-15-next-generation-speculative-decoding-dflash-v2

📰 OpenAI - OpenAI Merchant Feed Terms of Service

https://openai.com/policies/merchant-feed-terms-of-service


📰 OpenAI - Introducing the OpenAI Partner Network

https://openai.com/index/introducing-openai-partner-network


🔓 OpenAI - Training to cycle across Antarctica with ChatGPT

https://openai.com/index/cycling-across-antarctica

📰 NVIDIA - Boosting MoE Training Throughput with Advanced Fusion Kernels
Mixture-of-experts (MoE) models have quickly become a foundational component of modern, large-scale AI systems. They are widely adopted because they enable…

https://developer.nvidia.com/blog/boosting-moe-training-throughput-with-advanced-fusion-kernels/


📰 NVIDIA - Pretrained to Imagine, Fine-Tuned to Act: The Rise of World-Action Models
Quick glossary for readers new to VLA/WAM terminology VLA Vision-Language-Action model: a robot policy that starts from a pretrained VLM backbone and adapts it…

https://developer.nvidia.com/blog/pretrained-to-imagine-fine-tuned-to-act-the-rise-of-world-action-models/

🔓 [HF Models] inclusionAI - Ling-2.6-flash-base

https://huggingface.co/inclusionAI/Ling-2.6-flash-base


🔓 [HF Models] inclusionAI - Ling-2.6-1T-base

https://huggingface.co/inclusionAI/Ling-2.6-1T-base

🆕 [HF Models] zai-org - GLM-5.2


https://huggingface.co/zai-org/GLM-5.2

📰 Google AI Blog - Unlocking the Power of the TPU Stack: Introducing our new Developer Hub
Google has officially launched the TPU Developer Hub, a centralized educational resource designed to help model builders and developers maximize the performance of Google Cloud TPUs. The hub offers code-first resources, open-source recipes, and deep-dive documentation covering hardware architecture, software optimization, debugging, parallelism, and networking. These materials are tailored for both human developers and AI-assisted tools to streamline everything from large-scale training to low-latency inference workloads.

https://developers.googleblog.com/en/unlocking-the-power-of-the-tpu-stack-introducing-our-new-developer-hub/

📰 Claude Blog - Meet the winners of the Built with Opus 4.7 Claude Code hackathon


https://claude.com/blog/meet-the-winners-of-built-with-opus-4-7-claude-code-hackathon

📰 OpenAI - Predicting model behavior before release by simulating deployment


https://openai.com/index/deployment-simulation