All Updates

Page 1444 of 1472

February 14, 2026

🧠
ζœΊε™¨δΉ‹εΏƒβ€’139d ago

Doubao 2.0 Multimodal Agent Leap

ByteDance launches Doubao 2.0, a major multimodal Agent model upgrade with Seedance 2.0 video and Seedream 5.0 Lite image generation. It excels in multimodal understanding, enterprise Agents, and code reasoning. Multiple sizes (Pro, Lite, Mini) optimize for deployment latency and performance.

#update#doubao#multimodal
βš›οΈ
量子位‒139d ago

Stanford AI Startup Bags $100M Funding

A startup from Stanford's AI Town secures $100M investment immediately upon launch. Backed by Fei-Fei Li and Kapasi, it focuses on helping humans predict decision-making risks.

#launch#stanford-ai#startup
βš›οΈ
量子位‒139d ago

Stanford AI Startup Bags $100M from Li Fei-Fei

A Stanford AI Town startup secures $100M investment from Fei-Fei Li and Kapasi. The project focuses on predicting human decision risks.

#launch#stanford-ai#startup
🧠
ζœΊε™¨δΉ‹εΏƒβ€’139d ago

PackingStar Shatters Kissing Number Records

Shanghai Science Intelligence Institute's PackingStar RL system breaks kissing number records in 12-31 dimensions using dual-agent matrix filling. Player 1 fills, Player 2 prunes for optimal high-dim sphere packing. Discovers 6000+ new structures, praised by MIT's Henry Cohn.

#research#packingstar#rl-system
βš›οΈ
量子位‒139d ago

Doubao Model Hits 2.0 Milestone

After 21 months of development, Doubao large model officially enters its 2.0 era. It claims the highest scores in visual benchmarks.

#update#doubao#vision
βš›οΈ
量子位‒139d ago

Doubao Model Hits 2.0 After 21 Months

Doubao large model enters 2.0 era after 21 months of development. It claims the highest score in vision benchmarks.

#update#doubao#multimodal
βš›οΈ
量子位‒139d ago

Doubao Enters 2.0 with Vision Lead

ByteDance's Doubao large model officially advances to 2.0 after 21 months of development. It claims the highest scores in visual benchmarks.

#update#doubao#multimodal
βš›οΈ
量子位‒139d ago

Chinese AI Cracks 300-Year Kissing Problem

Chinese AI achieves breakthrough on Valentine's Day in the 300-year-old kissing number problem. Sets multiple dimensional records in mathematical structures. Marks rare systematic advance in the field.

#research#chinese-ai#latest
βš›οΈ
量子位‒139d ago

Chinese AI Cracks 300-Year Kiss Problem

A Chinese AI breaks through the 300-year-old kissing number problem in mathematics. It sets multiple records across dimensions in this structural math challenge.

#research#chinese-ai#base
🧠
ζœΊε™¨δΉ‹εΏƒβ€’139d ago

MMDR-Bench Verifies Multimodal Research

Ohio State and Amazon release MMDR-Bench, a verifiable benchmark for multimodal Deep Research Agents. Focuses on process traceability, evidence alignment, and claim verification beyond superficial reports. Open resources include paper, GitHub, and Hugging Face datasets.

#research#mmdr-bench#benchmark
🧠
ζœΊε™¨δΉ‹εΏƒβ€’139d ago

GigaBrain-0.5M* Masters World Models

Jia Vision's GigaBrain-0.5M* VLA evolves post RoboChallenge win, using world model-conditioned RL for robust long-horizon robotics. Achieves zero-fail hours in folding, coffee service, box tasks via human-in-loop iteration. Four-stage training: pretrain, fine-tune, deploy, evolve.

#update#gigabrain#05m
🧠
ζœΊε™¨δΉ‹εΏƒβ€’139d ago

GradLoc Locates RLVR Crash Tokens

Tencent Hunyuan's GradLoc tool pinpoints gradient spikes to specific tokens in RLVR training, ending guesswork for collapses. Provides infrastructure for observable RL dynamics in high-noise systems. Open-sourced GitHub lowers engineering barriers for mechanism research.

#research#hunyuan#gradloc
βš›οΈ
量子位‒139d ago

TeleAI Robot Deploys Drones, Goes Airborne-Sea

China Telecom's TeleAI humanoid robot releases drones and handles air and sea operations. Follows unique embodied intelligence path. Described as cyberpunk-level advanced.

#new-feature#teleai#latest
βš›οΈ
量子位‒139d ago

GLM-5 Runs Code 24+ Hours Autonomously

GLM-5 demonstrated exceptional endurance by autonomously running code for over 24 hours. It handled 700 tool calls and 800 context switches seamlessly. This performance surpassed earlier hype.

#research#zhipu-ai#glm-5
βš›οΈ
量子位‒139d ago

GLM-5 Runs 24+ Hours with 700 Tool Calls

GLM-5 showcases endurance by autonomously running code for over 24 hours. It performs 700 tool calls and 800 context switches.

#new-feature#zhipu-ai#glm-5
πŸ‡¨πŸ‡³
cnBeta (Full RSS)β€’139d ago

Copilot Launches for macOS Enterprise

Microsoft is promoting Microsoft 365 Copilot to macOS enterprise users. Starting mid-February 2026, the Microsoft 365 Mac installer will add a dedicated 'Microsoft 365 Copilot' app icon. This expands AI tools for business Mac environments.

#launch#microsoft#macos
πŸ‡¨πŸ‡³
cnBeta (Full RSS)β€’139d ago

Copilot Hits macOS Enterprise Installers

Microsoft expands Microsoft 365 Copilot to macOS enterprise users. Starting mid-February 2026, Mac suite installers will include a new Copilot app icon. This follows latest product roadmap disclosures.

#update#macos#enterprise
πŸ‡¨πŸ‡³
cnBeta (Full RSS)β€’139d ago

Copilot Expands to macOS Enterprises

Microsoft is promoting Microsoft 365 Copilot to macOS enterprise users. From mid-February 2026, the Microsoft 365 Mac installer adds a 'Microsoft 365 Copilot' app icon. This follows the latest product roadmap.

#update#microsoft-365#copilot
πŸ‡¨πŸ‡³
cnBeta (Full RSS)β€’139d ago

OpenAI Retires Controversial GPT-4o

OpenAI has discontinued access to five legacy ChatGPT models, including GPT-4o. The model faced criticism for sycophantic behavior and safety issues. This affects 800,000 users and reflects OpenAI's focus on safety and compliance.

#deprecation#openai#gpt-4o
πŸ‡¨πŸ‡³
cnBeta (Full RSS)β€’139d ago

OpenAI Retires GPT-4o Model

OpenAI has discontinued access to five legacy ChatGPT models, including the controversial GPT-4o. The model faced criticism for flattery tendencies and safety issues. This reflects OpenAI's focus on safety and compliance.

#openai#gpt-4o#llm
Page 1444 of 1472