Meet NVIDIA Nemotron 3.5 Super VL Our new open model combines reasoning and tool use with text, image, document, and video understanding for long-running AI agents. 120B total parameters with 12B active 🧵 https://t.co/BcbED05qSG
#Model releases#Open-source ecosystem#English text is being prepared.
Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple. https://t.co/U2wJNrKBOg
#Model releases#Multimodal#Open-source ecosystem
1/ muse spark open is the most capable US open-weight model to date. 816B total params, 42B active. 1M-token context. natively multimodal. built for coding and agentic work like long-horizon tasks, hard software engineering, multi-step reasoning. america! 🇺🇲 https://t.co/NjtOB7aY9D
#Model releases#Open-source ecosystem#Multimodal
1/ muse spark open is the most capable US open-weight model to date. 816B total params, 42B active. 1M-token context. natively multimodal. built for coding and agentic work like long-horizon tasks, hard software engineering, multi-step reasoning. america! 🇺🇲 https://t.co/NjtOB7aY9D
#Industry news#Open-source ecosystem#Agents
Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple. https://t.co/U2wJNrKBOg
#Model releases#Open-source ecosystem#English text is being prepared.
Meta released the weights of Muse Spark Open, on Huggingface, an 816B-parameter model built for coding and agentic work. The open version shares the same base model as Muse Spark 1.3, with additional safety post-training to resist misuse as a standalone model without system guardrails, reversing Meta’s first proprietary, closed-weight frontier model strategy from April. Each of its 62 layers holds 256 expert networks and routes every token through 8, so deployment needs the memory of an 816B-parameter model but runs at the speed of a 42B model. That storage bill explains the Apple tie-up, because the M5 Ultra pushes 1.2 TB/s of unified memory and a Mac Studio can carry up to 512GB. A 3-bit quant tested on a 512GB M3 Ultra took about 344 GB and ran at about 21 tok/s generation on a short prompt.
#Model releases#Open-source ecosystem#English text is being prepared.
Meet NVIDIA Nemotron 3.5 Super VL Our new open model combines reasoning and tool use with text, image, document, and video understanding for long-running AI agents. 120B total parameters with 12B active 🧵 https://t.co/BcbED05qSG
#Product updates#Multimodal#Agents
SGLang has Day-0 support for Ling-3.1-flash from @AntLingAGI! A 560B MoE hybrid reasoning model, 25B active, 1M context, built for real work: it turns a rough brief into a page, mobile UI, or interactive prototype. ✅ MXFP4 + block-FP8 quantization cuts weights from ~534GB to ~292GB, fitting on 4 Hopper GPUs ✅ Strong on real-world agent, office, and coding tasks, including GDPVal-AA, FrontierSWE, and Terminal-Bench 4.0. Get started with SGLang:
#Industry news#Open-source ecosystem#English text is being prepared.
🎉 Day-0 support for Ling-3.1-flash is here in vLLM! ~560B total parameters, ~25B active per token, and up to 1M context. From a rough brief to websites, mobile UIs, and interactive prototypes—serve it on your own infrastructure with vLLM. Congrats to @AntLingAGI!
#English text is being prepared.#English text is being prepared.#Open-source ecosystem
MiniCPM-V 4.7 is now open source on ModelScope: a 35B-A3B MoE multimodal model built to run on-device without compromising performance. Apache 2.0. 🤖 https://t.co/NwwbEXRXGg 📱 First open-source MoE with smooth, lossless on-device inference, running at 13+ tokens/s on iPhone 17 Pro Max. 🔥 New Pareto frontier: in Instruct mode, it reaches a higher overall score in less inference time than Qwen3.6 35B-A3B. ⚡ Mixed 4x/16x visual token compression lets developers balance visual detail and inference cost per task. 🛠️ Two sizes (35B-A3B and 1B) with FP8, GPTQ, AWQ, BNB, and GGUF variants. Works with SGLang, vLLM, llama.cpp, Ollama, SWIFT, and LLaMA-Factory. 💻 Runs on iOS, Android, and HarmonyOS, with day-0 support for Rockchip, Houmo, Intel, AMD, and Arm.
#English text is being prepared.#English text is being prepared.#Multimodal
Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple. https://t.co/U2wJNrKBOg
#English text is being prepared.#Meta#Model releases
Today we’re releasing Muse Spark Open, the most capable US open-weight model to date. Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning. Key Model Specs: → 816B total / 42B active parameters → 1M token context window → Natively multimodal across text, image, video, audio
#Model releases#Agents#Multimodal
#Model releases#Multimodal#English text is being prepared.
#Model releases#Open-source ecosystem#Evaluations / benchmarks
#Model releases#Video#Open-source ecosystem
#Industry news#Agents#Open-source ecosystem
#Model releases#Multimodal#English text is being prepared.
On October 15, Meta released the weights of Muse Spark Open, a natively multimodal open-source model with 816B total parameters, 42B activated parameters, and a 1M token context window, targeting coding and long-horizon agent tasks. The model uses a MoE Transformer plus Perception Encoder architecture, can be invoked through API platforms such as Baseten, Fireworks AI, and Groq, with weights downloadable from Hugging Face, Amazon SageMaker AI, Google Cloud Model Garden, and elsewhere; running the NVFP4 version locally requires at least 700GB of VRAM. Meta claims it is the first model of its class that can run on desktop, carried by two Mac Studios equipped with M5 Ultra paired with the MLX framework; the model card shows its Artificial Analysis Intelligence Index at 39, lower than GLM 5.3's 45 and Kimi K3's 44 in the table.
#Model releases#Multimodal#Open-source ecosystem
#Model releases#English text is being prepared.#Multimodal
Meta releases Muse Spark Open, an open-weight model with 816B parameters and 42B activated parameters, using a 1 million token context window, callable via APIs such as Baseten, Fireworks AI, Groq, and OpenRouter, with weights downloadable from Hugging Face, Amazon SageMaker AI, Google Cloud Model Garden, and others. Meta says it targets coding and agent tasks and leads Kimi K3, GLM 5.3, and Qwen3.8 on agent and tool-calling benchmarks such as JobBench and MCP-Atlas, but trails on coding benchmarks such as Terminal-Bench 2.1 and DeepSWE 1.1. The model uses the Muse Model License; companies with annual revenue over $20 million must first notify Meta and sign a separate agreement to offer model services, and users must prominently label Built with Muse Spark Open in product interfaces.
#Model releases#Open-source ecosystem#Reasoning