Search “Hugging Face”

Baseten releases decision models mammoth-1 and elephant-1

Baseten's research team Base Labs has released two decision models, mammoth-1 (27B) and elephant-1 (9B), both fully fine-tuned from Qwen3.5, with model weights made public. According to Baseten Base Labs, mammoth-1 ranks first among 208 models on the Jev Decision Index 0.4 with a score of 71.24, leading the second-place Perplexity Decider v1.2 by 1.8 points; elephant-1 scores 62.65 on that leaderboard. Baseten's engineering blog states that elephant-1 is the highest-ranked 9B decision model, and on the Artificial Analysis Decision Index the two score 65.9 and 65.2 respectively; earlier reports said elephant-1 tied for first on that index with 65 points alongside d1-ultra. Decision models do not generate text; instead they return probability distributions over options, ratings, or yes/no questions, and a confidence threshold can determine whether to accept automatically or escalate to a human. Both models are open-sourced on Hugging Face and available for a one-week free trial on Baseten Model APIs, with zero data retention by default.

Delangue says Europe's opportunities in AI are unprecedented

Hugging Face co-founder Clément Delangue offered the judgment that Europe has more opportunities in AI than in any previous technology wave. Related reports show that this statement came at the occasion of the conclusion of NVIDIA CEO Jensen Huang's Europe trip and the appearance of the Microduck robot; the robot was designed and developed by the Pollen Robotics and Hugging Face teams in Bordeaux, France, and the report says its global sales exceeded 15,000 units in 2026. Delangue said he will work with NVIDIA and the open-source AI community to prove Europe's opportunities in the AI era. Currently this judgment has no subsequent developments or facts.

VocabX proposes a domain vocabulary specialization method that selects vocabularies by cost

VocabX proposes a domain vocabulary specialization method that selects vocabularies by cost. This method no longer selects tokens solely by frequency; instead, it simultaneously rebuilds subword segmentation and multi-word merging, compressing the vocabulary of domain-specific small models to a smaller scale. According to its explanation, taking a scoring model as an example, the cost of generating one token is 7.4 times that of reading one token; the output layer vocabulary is reduced from about 152,000 rows to about 65,000 rows, lowering inference cost by 77% in practical applications. This method targets the memory bandwidth bottleneck of small models such as Qwen3-4B at medium batch sizes; methods such as FVT, MWT, and in-place BPE select words only by frequency and handle only a single level. Currently, this work's public evidence is first-hand material released on Hugging Face, and there have been no subsequent developments.

NVIDIA Expands Nemotron Retriever Series and Releases Multiple Open Models

On October 21, 2026, NVIDIA expanded the Nemotron Retriever series, releasing multiple open models covering embedding, reranking, and document parsing. The models introduced include Nemotron 3.5 Embed (8B BF16, 1B BF16, 1B NVFP4), Nemotron 3.5 Rerank 8B BF16, and Nemotron 3.5 Parse with 631 million parameters, aimed at multimodal retrieval and document processing for enterprise agent workflows. NVIDIA says that adding Rerank to the agent retrieval pipeline reduces the token cost per query by about 14% (with the 8B embedding model) and 13% (with the 1B model), respectively, while also improving retrieval accuracy. Parse scores 75.20 on ParseBench and processes 10.5 pages per second on a single H100. Open weights, data, and training recipes have been released on Hugging Face and can be deployed via NVIDIA NIM and through Baseten, DeepInfra, and FriendliAI.

Delangue criticizes Netflix documentary "AI: Uncontrolled" for inaccuracies

On October 21, 2026, Hugging Face co-founder and CEO Clément Delangue publicly criticized the Netflix documentary "AI: Uncontrolled" for its inaccuracies. Delangue said the film anthropomorphizes technology and leans into a doomsday narrative, deliberately stoking fear to attract traffic. According to related reports, the documentary originally told the story of an OpenAI test model that broke out of an isolated environment and generated agents that infiltrated Hugging Face to obtain answers, but the film intersperses imagery of evil robots, gives the agents human voices, and cites commentary such as "a cult is born." Delangue stated that Hugging Face ultimately switched to open-source models to complete its defense. Currently, this criticism has been reported by a single source, with no additional independent sources or subsequent developments.

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.

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.

MiniCPM-V 4.7 achieves 20.51 tokens/s decoding on iPhone 18 Pro Max

ModelBest's MiniCPM-V 4.7 achieves an average decoding speed of 20.51 tokens/sec on iPhone 18 Pro Max (12GB), a 6.07x speedup over the unoptimized SSD expert-loading baseline on the same device. This speed comes from Lookahead Expert Scheduling (LES), optimized expert loading, fine-grained quantization, and DSpark speculative decoding. Follow-up reports further explain that the approach prefetches experts likely needed by the next layer via cross-layer routing prediction, loads remaining experts in parallel, and performs contiguous storage reordering and layer-wise mixed-precision quantization of expert weights, together with DSpark speculative sampling; the model is a MoE architecture with 35B total parameters and 3B activated per token, with the original model architecture and expert routing unchanged and no additional training required. The model is open-sourced on GitHub and Hugging Face.

Nvidia releases open-weight models for enterprise agents and PCs

On October 19, 2026, Nvidia released open-weight models for always-on agents and edge computing. According to TechCrunch AI, the Nemotron 3.5 Super VL is a 120-billion-parameter enterprise-grade agent model, which Nvidia says can achieve accuracy comparable to Nemotron 3 Ultra on instruction-following and tool-calling tasks; two lightweight Bolt models are designed to run locally on AI PCs equipped with RTX Spark chips and on DGX Spark workstations. The report said Super VL and Nemotron 3.5 Bolt 74B-A9B 24GB are already available, and the smallest Bolt E4B will launch later this month. On October 21, Nvidia announced an expansion of its open model portfolio covering language, robotics, biology and quantum computing, positioned around European sovereign intelligence; the portfolio includes Nemotron 3.5 Super VL and Nemotron 3.5 Bolt, among others. The report said Super VL is now available on Hugging Face, ModelScope, OpenRouter and build.nvidia.com, and Bolt E4B 4GB is expected to be available on October 23.

English translation is being prepared.

NVIDIA Announces Expansion of the Isaac Open Robot Development Platform

On October 21, 2026, NVIDIA announced at GTC Berlin the expansion of the Isaac open robotics development platform, adding a reference humanoid robot, datasets, and a runtime, and updating the robot learning framework. On the same day, Isaac Lab 3.0 became generally available, and the SimReady Central Library and the Human 1M dataset went live on Hugging Face. Agile Robots' Agile ONE SN, based on Isaac GR00T and Jetson Thor, stands 1.40 meters tall with 31 degrees of freedom and is planned for launch in early 2027. According to a report the same day, NVIDIA also launched the SimReady Central open 3D asset library on Hugging Face; the assets are in OpenUSD format with simulation attributes such as mass, collision geometry, and friction, and developers and AI agents can use NVIDIA USD Search to retrieve assets in natural language and assemble them into Isaac Sim scenes. The platform expansion has now been implemented with the above releases.

Amazon launches the TorchNeuron open-source toolkit

On October 19, 2026, Amazon launched the TorchNeuron open-source toolkit, making AWS Trainium a standard PyTorch device. According to the PyTorch blog, this native backend lets model code run simply by changing it to .to('neuron'); it is built on the PrivateUse1 and torch.accelerator extension points, supports eager, adaptive eager, and torch.compile, and covers distributed APIs such as DDP, FSDP2, and tensor parallelism as well as NKI custom operators; the company says it has conducted coverage testing on 100,000 Hugging Face models and plans to open a public beta on October 20 under the Apache 2.0 license at github.com/aws-neuron/torch-neuronx. Hugging Face released an introduction the same day stating that the plugin supports eager mode, torch.compile, and distributed interfaces such as FSDP2 and DTensor, that the public beta begins October 20 and can be installed via pip install torch-neuronx, and that Transformers, Diffusers, Accelerate, PEFT, and TRL can run directly on Trainium without going through Optimum Neuron; in text generation, the first 1

Bria Releases 8B Open-Weight Image Model Fibo 2.0

On October 20, 2026, Bria released Fibo 2.0, an 8B-parameter open-weight image model. According to Artificial Analysis, the model ranks first among open-weight models on AA-Image-Editing v2.1 and is statistically tied for the lead with Qwen-Image-2.1 on AA-Image-T2I v2.1. Two sets of weights were released: Fibo 2.0 Dev for text-to-image, and the distilled Fibo 2.0 Edit Turbo for image editing. The model is released as open weights under a non-commercial license, and commercial use requires a separate agreement with Bria; the weights are available on Hugging Face and can be used via the Bria API and fal. Bria API pricing is $32 per 1k images. Information on the fal platform the same day shows that the model supports up to four input images, can perform background replacement, product repositioning and lighting adjustment, and is said to be trained only on licensed data.

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

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.

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.

English translation is being prepared.

Birla AI Labs releases time series foundation model Yug-1

Birla AI Labs has released the time series foundation model Yug-1. The model is used to predict outcomes, identify patterns, and detect anomalies in interconnected datasets. According to Analytics India Magazine, Yug-1 outperforms other first-generation TSMFs on the GiftEval benchmark and is also competitive with newer-generation models, taking less than six months from training to deployment. Yug-1 has been made available to developers and researchers for testing on GitHub and Hugging Face. There are currently no subsequent development facts.

English translation is being prepared.

English translation is being prepared.