←Search results

Search “Hugging Face”

By relevance · Total 342 results Clear search

Friday

  1. Story5 reports · 1 facts

    SGLang natively supports and powers the Beam production API

    On October 16, 2026, LMSYS announced that SGLang now natively supports Reflection's Beam model and is being used to power its production API. Beam is a 501B-parameter MoE model with 23B active parameters per token and a native 1M-token context window, planned for release under Apache 2.0; SGLang provides BF16, NVFP4, and dynamic FP8 support for it, along with DFlash2 speculative decoding and hybrid attention prefix caching. Running OpenHands multi-turn agentic coding replay with NVFP4 on 4 NVIDIA GB300 GPUs, it reached 127.6K tokens/sec per GPU and 185.87 tokens/sec per user (p50) at a concurrency of 32. On the same day, Reflection released the Beam model weights, with Ollama offering them free for a limited time as a day-zero partner with zero data retention; Beam has approximately 501 billion total parameters, about 23 billion active parameters per token, a 1 million-token context window, and is released under the Apache 2.0 license. Subsequent information from AWS states that the Beam weights have been released on Hugging Face, offering three checkpoints—BF16, FP8, and NVFP4—and are now available on Amazon SageMaker JumpStart, and can also be accessed via SageMa

    a offering them free for a limited time as a day-zero partner with zero data retention; Beam has approximately 501 billion total parameters, about 23 billion active parameters per token, a 1 million-token context window, and is released under the Apache 2.0 license. Subsequent information from AWS states that the Beam weights have been released on Hugging Face, offering three checkpoints—BF16, FP8, and NVFP4—and are now available on Amazon SageMaker JumpStart, and can also be accessed via SageMa

    Matched fieldsSummary

  2. Story32 reports · 3 facts

    Meta releases open-weight model Muse Spark Open

    On October 15, 2026, Meta released the open-weight model Muse Spark Open, with the weights going live on Hugging Face the same day under the Muse Model License. According to the PyTorch Blog, the model has 816B total parameters, 42B activated parameters, and a 1M token context window; it is natively multimodal and aimed at coding and agentic tasks, adds safety post-training on top of Muse Spark, and can run on SGLang and vLLM. On October 16, the model launched on Amazon SageMaker JumpStart with two variants, BF16 and NVFP4; the same day it entered live testing in Arena's Agent Arena and launched in Code Arena's WebDev and other projects. On October 17, Arena announced that Muse Spark Open (Max) ranked first among US open models on the Code Arena WebDev leaderboard with a score of 1511, placing 13th among all open models.

    On October 15, 2026, Meta released the open-weight model Muse Spark Open, with the weights going live on Hugging Face the same day under the Muse Model License. According to the PyTorch Blog, the model has 816B total parameters, 42B activated parameters, and a 1M token context window; it is natively multimodal and aimed at coding and agentic tasks, adds safety post-training on top of Muse Spark, and can run on SGLang and vLLM. On October 16, the model launched on Amazon SageMaker JumpStart with

    Matched fieldsSummary

  3. Story16 reports · 3 facts

    Nvidia releases Nemotron 3.5 Super VL and Bolt models

    On October 15, 2026, Nvidia announced the launch of Nemotron 3.5 Super VL and the edge-oriented 3.5 Bolt models. According to Nvidia's official introduction, Super VL is a 120B-parameter open model aimed at always-on agent tasks, and the company says it has the highest throughput among comparable open models, 5 times faster than the next-best model; it also launched two edge models, Bolt 74B-A9B 24GB and Bolt E4B 4GB, with the former able to run agent tasks locally on RTX/DGX Spark and the latter aimed at voice and conversational agents on GeForce RTX and Jetson. Super VL and Bolt 74B-A9B are already available on Hugging Face, and Bolt E4B is scheduled for release on October 23. On the same day, SGLang announced Day-0 support for Super VL, reaching 243.9 tok/s on a single RTX 6000 PRO with the help of MTP; Prime Intellect and Nvidia used GRPO to post-train it for about 11 hours on 128 GB200 GPUs, raising its score on the public AutomationBench-AA set from 34.0% to 57.7%. That evening OpenRouter listed the model, and the next day Perplexity announced that Pro and Max subscribers and the Agent API can use it.

    e to run agent tasks locally on RTX/DGX Spark and the latter aimed at voice and conversational agents on GeForce RTX and Jetson. Super VL and Bolt 74B-A9B are already available on Hugging Face, and Bolt E4B is scheduled for release on October 23. On the same day, SGLang announced Day-0 support for Super VL, reaching 243.9 tok/s on a single RTX 6000 PRO with the help of MTP; Prime Intellect and Nvidia used GRPO to post-train it for about 11 hours on 128 GB200 GPUs, raising its score on the public

    Matched fieldsSummary

  4. Story2 reports · 1 facts

    StepFun updates Step Code to v0.1.4

    On October 15, 2026, StepFun updated Step Code, the open-source agent harness built around Step 5 Preview, to v0.1.4, helping agents track goals in long and complex tasks, keep making progress, and check against verifiable results whether tasks are truly completed. The company says that since Step Code was open-sourced, four versions have been released, roughly once every 3 working days, driven mainly by community feedback. On the same day, StepFun officially opened the Step 5 Preview model weights, with BF16 and FP8 versions launched simultaneously on Hugging Face and ModelScope; the model uses a sparse MoE architecture, with 600B total parameters and 27B activated parameters, and supports a 1 million Token context window.

    ether tasks are truly completed. The company says that since Step Code was open-sourced, four versions have been released, roughly once every 3 working days, driven mainly by community feedback. On the same day, StepFun officially opened the Step 5 Preview model weights, with BF16 and FP8 versions launched simultaneously on Hugging Face and ModelScope; the model uses a sparse MoE architecture, with 600B total parameters and 27B activated parameters, and supports a 1 million Token context window.

    Matched fieldsSummary

Sunday

  1. Story1 reports · 1 facts

    Perplexity Decider launches on Vercel AI Gateway

    On October 17, 2026, Perplexity's Decider V1.2 decision model launched on Vercel AI Gateway. Vercel developers can now use perplexity/pplx-decider-v1.2 for complex agentic workflows and perplexity/pplx-decider-flash-v1.2 for ultra-fast decisions. Aravind Srinivas reposted the news. Earlier the same day, reports said Perplexity had open-sourced two decision models, pplx-decider-flash-v1.2 and pplx-decider-v1.2: the former has 4B parameters and costs 1 cent per million tokens, and is called the strongest 4B model on the HF Decision Index; the latter has 27B parameters and costs 2 cents per million tokens, and ranks 2nd on the overall HF Decision Index leaderboard. Both are now available on the Perplexity Decisions API and offer open weights on Hugging Face. Perplexity Developers said this generation of open-weight multimodal decision models improves both text and vision scores, delivers inference up to 5 times faster, and completes all questions in a single request in one forward pass.

    strongest 4B model on the HF Decision Index; the latter has 27B parameters and costs 2 cents per million tokens, and ranks 2nd on the overall HF Decision Index leaderboard. Both are now available on the Perplexity Decisions API and offer open weights on Hugging Face. Perplexity Developers said this generation of open-weight multimodal decision models improves both text and vision scores, delivers inference up to 5 times faster, and completes all questions in a single request in one forward pass.

    Matched fieldsSummary

  2. Story1 reports · 1 facts

    Reflection announces a $10,000 Hub credit for the Beam project

    Reflection announces a $10,000 Hub credit for the Beam project. The news was released on October 16, 2026. At the same time, Hugging Face released the Beam model: a 501B-parameter sparse MoE open-source model aimed at long-context reasoning and agent tasks with a 1M context, activating only about 23B parameters per token. The model uses 52 layers of hybrid attention, of which 43 layers use a 512-token sliding window and 9 layers use global attention; the KV cache grows by 36 KiB per token, about one-sixth that of a full-attention scheme. Official checkpoints are provided in BF16, FP8, and NVFP4; NVFP4 can compress the weights from about 1 TB to about 280 GB, and day-0 support is already available in transformers, vLLM, and SGLang. There are currently no subsequent developments.

    Reflection announces a $10,000 Hub credit for the Beam project. The news was released on October 16, 2026. At the same time, Hugging Face released the Beam model: a 501B-parameter sparse MoE open-source model aimed at long-context reasoning and agent tasks with a 1M context, activating only about 23B parameters per token. The model uses 52 layers of hybrid attention, of which 43 layers use a 512-token sliding window and 9 layers use global attention; the KV cache grows by 36 KiB per token, about

    Matched fieldsSummary

Thursday

Wednesday

  1. Story2 reports · 1 facts

    Liquid AI Releases LFM2.5-24B-A2B On-Device Reasoning Model

    On October 13, 2026, Liquid AI released the LFM2.5-24B-A2B on-device agent model. With 24B total parameters and only 2B activated per token, it is the largest and final model in the LFM2.5 family, and both the base and post-trained versions have open weights on Hugging Face and Playground. According to the official blog, it uses a pure reasoning mode with explicit chain-of-thought, expands the context window from 32,768 to 131,072 tokens, grows the vocabulary from 65K to 128K, and scales pretraining data from 17T to 33T tokens; it leads or ties Qwen3.6-35B-A3B on five of eight agent and tool-calling benchmarks, scoring 27.69 versus 8.72 on AutomationBench and 48.09 versus 38.19 on LiveClaw; it decodes at 184 tokens/s on an Apple M5 Max and outputs over 21K tokens/s under high concurrency on a single NVIDIA B300 GPU. On October 14, OpenRouter announced that the model is now available, saying it reasons before answering and targets agent and tool-calling scenarios.

    With 24B total parameters and only 2B activated per token, it is the largest and final model in the LFM2.5 family, and both the base and post-trained versions have open weights on Hugging Face and Playground. According to the official blog, it uses a pure reasoning mode with explicit chain-of-thought, expands the context window from 32,768 to 131,072 tokens, grows the vocabulary from 65K to 128K, and scales pretraining data from 17T to 33T tokens; it leads or ties Qwen3.6-35B-A3B on five of eigh

    Matched fieldsSummary

Monday

Sunday

  3. Story3 reports · 1 facts

    Qwen Releases Qwen-Image-2.1-Turbo Open Weights

    On October 9, 2026, Qwen released Qwen-Image-2.1-Turbo and made the model's weights openly available. According to the official description, the model is based on Qwen-Image-2.1 and uses the same 7B visual generation architecture; it requires only 8 denoising steps to generate and edit images. The weights are now available on ModelScope and Hugging Face. This version can still generate 2K images from text and supports continued creation through natural language editing, such as adding accessories or changing scenes. Using Diffusers to load QwenImage21Pipeline allows running with the recommended 8-step sampling scheme. The APIs for Qwen-Image-2.1 Pro and Turbo have also officially launched. There are currently no subsequent developments.

    Qwen-Image-2.1 and uses the same 7B visual generation architecture; it requires only 8 denoising steps to generate and edit images. The weights are now available on ModelScope and Hugging Face. This version can still generate 2K images from text and supports continued creation through natural language editing, such as adding accessories or changing scenes. Using Diffusers to load QwenImage21Pipeline allows running with the recommended 8-step sampling scheme. The APIs for Qwen-Image-2.1 Pro and T

    Matched fieldsSummary