Ricursive发布Ripple-65B Preview开源模型，称5人团队以3200万美元完成训练
Ricursive Intelligence发布开源模型Ripple-65B Preview，称其在Terminal-Bench 2.1编程测试中得分80.5%，模型权重采用Apache 2.0许可证。
Loading…
bibibi.ccbibibi.ccLoading…
Connect scattered updates into a story you can understand.
Ricursive Intelligence发布开源模型Ripple-65B Preview，称其在Terminal-Bench 2.1编程测试中得分80.5%，模型权重采用Apache 2.0许可证。
Ricursive Intelligence称发布从零预训练的65B稠密开源权重模型Ripple-65B Preview，并披露其在TerminalBench 2.1的测试成绩及上线渠道。
Beating AI reports that Mistral has open-sourced improved Large 4(Le Chonk) weights, providing FP8 and NVFP4 versions. The report also discloses information on its programming, efficiency, and deployment licensing.
Vitalik Buterin When replying to Ruxandra Teslo, said that AI although extremely knowledgeable and adept at identifying references, lacks wisdom and can make only superficial judgments about a text's viewpoint.
DiffuSpace released and open-sourced a diffusion model based on a modification of Qwen3.6-35B-A3B Clover-35B-A3B-Preview. Its disclosed internal tests reportedly showed that the model reached GB300 on a single GPU 680 Token/s。
The Ant Group Ling team announced the open-sourcing of Ling-3.1-flash-1019 and the base model, which scored v4.3.2 on the Artificial Analysis Intelligence Index 44。
Inception announces the launch of the diffusion large language model Mercury 3, supporting 1000000 Token context; it is available through multiple platforms and compatible with OpenAI API。
ByteDance Research found that DeepSeek-V4 in long-text processing, answers periodically change as the length of irrelevant content varies, while V4.1-Flash the fluctuations narrow somewhat but persist.
A social post proposes a possible workflow for video generation.
According to reports, Alibaba has appointed senior AI researcher Liu Dayiheng to lead its Qwen large language model project, and the official Yunqi Conference website also lists his related position.
