此刻温度最高

48 小时内每个独立来源只算一次，热度每 24 小时减半。排在前面的，是真正有很多人在说的 8 件事。

2026·10·16

NO.01↓ 16%

Meta 发布 Muse Spark Open，支持 Mac 和云端；Artificial Analysis 发布基准测试结果

Meta released downloadable weights for Muse Spark Open, a multimodal model for coding and agentic work that Artificial Analysis ranks first among US open-weight models on its Intelligence Index, scoring 39 at the xhigh setting. The model has approximately 816 billion parameters, with 42 billion active per token, and supports up to one million tokens of context. It accepts text, images, audio and video and generates text. Artificial Analysis’s initial AA-AgentPerf testing repo…

24 小时热度峰值102 · 18 小时前

24 小时前现在

最新进展Artificial Analysis reports Muse Spark Open throughput results on AA-AgentPerf

vLLM · X、Artificial Analysis · X 等 12 个来源·12位参与者

热度指数

NO.02↓ 16%

Google 在全球发布 Gemini 4 Argon，扩大访问范围并报告其前 72 小时处理了 25 万亿个 token

Google launched Gemini 4 Argon globally through paid developer services and subscriptions, expanded access through Antigravity and third-party apps, and processed about 25 trillion tokens in its first 72 hours, according to Google’s Logan Kilpatrick. Google offers Argon through the paid Gemini API and Gemini platform, and to Pro and Ultra subscribers in the Gemini app and Search. Antigravity launched with Ultra access and is adding paid Pro subscribers through its desktop ap…

6个来源 · 6位参与者

NO.03↓ 16%

Black Forest Labs 发布 FLUX 3 Image Klein，开放模型权重并支持消费级 GPU

Black Forest Labs has released FLUX 3 Image Klein, a 5.6-billion-parameter open-weights model for image generation and editing, with launch-day support in Hugging Face Diffusers and vLLM-Omni. The Decoder reported that the model supports up to ten reference images and native 4K output, while its INT8 build runs on NVIDIA GPUs with 8 GB of VRAM and generates an image in under 9 seconds on an RTX 4070. The weights use the FLUX Kommunity License, which permits free commercial us…

4个来源 · 4位参与者

继续看No.04–10

参与者 · 24 小时走势 · 热度指数

04
Reflection 发布 Beam 1.0：501B 开放权重 MoE 模型新
Reflection launched Beam 1.0, a 501B mixture-of-experts model, making its weights available across platforms under the Apache 2.0 license in collaboration with more than 40 ecosystem partners. Reflection’s announcement lists a tokenizer, reference implementation, BF16, FP8 and NVFP4 weights, a DFlash2 drafter, and Mirror CLI. The Beam 1.0 NVFP4 checkpoint is available on Hugging Face. NVIDIA AI said Beam was trained on NVIDIA GB300 NVL72 and delivers more than 70K tokens per…
2个来源
19新上榜
05
Reflection发布Beam开放权重模型，发布首日即支持vLLM新
Reflection released its first model, Reflection-Beam-1.0, as an open weights model on Hugging Face, with launch-day serving support in vLLM. Artificial Analysis gave Beam a score of 30 on its Intelligence Index, placing it second among US open weights models. Beam has 501B total parameters, 23B active per token and a 1 million-token context window. It combines hybrid attention with Harmony-format reasoning and tool use. vLLM said it worked with Reflection and Inferact on ser…
2个来源
19新上榜
08
OpenAI推出Rosalind学术计划，为符合条件的美国大学实验室提供赞助访问权限发酵中
OpenAI has launched the Rosalind Academic Plan, offering eligible research labs at U.S. universities sponsored access to GPT-Rosalind and life sciences research tools. According to The Decoder, approved labs can receive up to 30 free Business Standard seats for up to twelve months; API access is excluded. The plan centers on GPT-Rosalind, a life sciences model that connects to specialized tools through Rosalind Workbench. OpenAI said the tools are intended to help researcher…
2个来源
15↑ 119%
09
Anthropic 向 METR 六名研究人员提供内部访问权限，以独立评估 AI 安全发酵中
Anthropic said it has onboarded six researchers from the nonprofit METR to independently assess its AI models, safety practices, and risk management, giving them internal access and freedom to publish without a company veto. The researchers receive company hardware, office space, and access to tools and systems roughly comparable to those available to Anthropic’s own risk assessment teams. Ars Technica reported that Anthropic is not paying METR for the arrangement. The publi…
2个来源
15↑ 126%
10
微软 AI 宣布自研 3 万亿参数级别的基础模型
穆斯塔法·苏莱曼表示，微软 AI 已完全自主构建了权重规模达 3 万亿参数的预训练基础模型，后训练结果尚待公布。苏莱曼称这些基础模型“处于业界领先水平”，并表示后训练结果很快会公布。
1个来源
10↓ 16%