Fu Sheng Unveils Orion-14B Large Model with 14 Billion Parameters

During the Fu Sheng 2024 Opening AI Grand Lecture and the Orion Star Large Model Launch Event, Fu Sheng unveiled the Orion Star Large Model (Orion-14B). This model boasts a parameter scale of 14 billion, covers common languages and professional terminology, and has achieved the best results on multiple third-party test sets. The Orion Star Large Model supports ultra-long texts, up to 320K tokens, with a reasoning speed of 31 tokens per second. It excels particularly in multilingual capabilities, especially in Japanese and Korean. To meet the application needs of enterprises, Orion Star also introduced fine-tuned models and several applications to help enterprises enhance operational efficiency and decision-making capabilities.

Zhipu AI's New Generation Large Model Roadmap Revealed: Parameter Count Exceeds Trillion, Dual-Track Technology Path Aims for AGI

Zhipu disclosed important technical signals of its new flagship large models GLM-5.4 and GLM-5.5: the parameter scale of both models has officially entered the trillions, and they will explore two new paths in the underlying technical architecture, seen as a step forward for domestic large models into the deep waters of AGI. The 'dual-track new path' has attracted widespread attention in the industry, marking Zhipu's advancement in the evolution of next-generation large model technology.

Alibaba Trials QwenBook AI Device, Team Size Around 150 People

The Alibaba Cloud Yingying team is developing the AI device QwenBook, positioned as a native intelligent agent computer, with a form similar to the Qwen tablet, still in the early trial phase, focusing on office scenarios, integrating the Qwen3.8-Max and Flash models. The product definition and hardware are self-developed, integrating the Yingying Cloud PC system and end-cloud capabilities, with a team of about 150 people.

Zhipu Launches GLM-5.3-FlashX: Speed Reaches 200 Tokens/s, Domestic Computing Power Advances Again

Zhipu AI launches GLM-5.3-FlashX, with the API also going live, offering a maximum output of 200 tokens/s. It focuses on intelligence, price, and speed, providing high-throughput, low-latency inference for enterprise developers. The predecessor, GLM-5.3-Flash, was previously introduced overseas under the name Ox Alpha. It gained popularity due to its strong intelligence and cost-effectiveness at the same size, with increasing usage volume. Zhipu is now supporting growing demand.

AntGroup Open Sources SingProbe: Real-Time Risk Blocking for Large Model Generation with Lightweight Probe, Computing Overhead Below 0.5%

Ant Group's AI Security Lab opens the internal security guardrail technology called SingProbe. Unlike external review, it reuses the hidden states of the base model for inference, achieving real-time token-level risk assessment with less than 0.5% additional computing power. It can simultaneously perform intent classification, safety detection, and hallucination recognition, and can block risks at the millisecond level before output in high-risk scenarios such as healthcare. It has been adapted to 29 mainstream large models.

Fu Sheng Unveils Orion-14B Large Model with 14 Billion Parameters

Related Recommendations

Zhipu AI's New Generation Large Model Roadmap Revealed: Parameter Count Exceeds Trillion, Dual-Track Technology Path Aims for AGI

Tencent Hunyuan brings the classic critical point size theory into large model reinforcement learning PPO, with throughput increased by up to 2.29 times and GRPO saving 29% training time

Alibaba Trials QwenBook AI Device, Team Size Around 150 People

Zhipu Launches GLM-5.3-FlashX: Speed Reaches 200 Tokens/s, Domestic Computing Power Advances Again

AntGroup Open Sources SingProbe: Real-Time Risk Blocking for Large Model Generation with Lightweight Probe, Computing Overhead Below 0.5%