2026-08-14
AI Briefings Daily Report.

How to Maximize Token Value in Claude Code Sessions ↗

Practical tips for how to run efficient sessions that get the most value from every token.

OpenAI and Anthropic in price war as Chinese AI rivals gain ground ↗

Privacy Center Currently, only residents from GDPR countries and certain US states can opt out of Tracking Technologies through our Consent Management Platform. Additional options regarding these technologies may be available on your device, browser, or through industry options like AdChoices. Please see our Privacy Policy for more information.

GPT-5.6 Sol launches Ultrafast mode, 14x faster inference, powered by Cerebras ↗

OpenAI is launching "Ultrafast," a new inference mode that delivers GPT-5.6 Sol at up to 750 output tokens per second, powered by Cerebras hardware from their $10 billion partnership. Together with "Standard" and "Fast," Ultrafast creates a three-tier pricing structure that turns inference speed into its own product.

DeepSeek peak-valley pricing update: V4-Pro officially released, API introduces peak-valley pricing ↗

We’re launching DeepSeek-V4-Pro today! 🚀

Samsung Health AI model analyzes wearable biosignal data ↗

Samsung Research America’s Digital Health Team has presented two AI foundation models designed to learn from wearable biosignals. The work centres on data captured by smartwatches, including heart activity, sleep, and physical activity.

New Google Chrome browser feature exposed: Gemini helps you modify weak and duplicate passwords ↗

Google Chrome Canary version is testing a new AI feature that lets Gemini automatically modify weak or duplicate passwords for you, making it more convenient than previously only reminding users.

Claude Code now runs daily maintenance on Anthropic's software with a 46 percent merge rate ↗

Anthropic is testing whether Claude Code can handle daily maintenance of the company's own apps, from crash fuzzing to dead-code removal. In a few weeks, the AI created 388 pull requests, and 46 percent were merged after human review. Claude Code inventor Boris Cherny sees this as "early signs of life that this might be possible."

Tencent Hunyuan senior researcher Xu Can transferred to WeChat WeLM, WeChat AI may enter acceleration phase ↗

According to 36Kr, Tencent Hunyuan senior researcher Xu Can has been transferred to the WeChat WeLM team, mainly responsible for large model R&D. He previously served as head of the post-training direction; this transfer occurred during WeChat's phase of strengthening self-developed large models and Agent capabilities.

Apple Collaborates with Alibaba to Train China-Specific AI Models ↗

The rare US-China partnership comes as Apple prepares to roll out its on-device generative AI service in China.

Zhipu releases GLM-5.3, calling it the strongest open-weight programming model ↗

Zhipu AI has released GLM-5.3, a model that, according to its own benchmarks, is the most powerful open-weights coding model, with a 50 percent improvement over its predecessor through post-training alone. Trained for cybersecurity, GLM-5.3 helped security teams find 2,436 vulnerabilities across 269 projects. The model weights are set to go open source in two weeks.

dots3-note Preview open-sourced: 280B parameter lightweight model, focused on long-horizon agents and multimodal reasoning ↗

Google AI health coach to integrate Abbott blood glucose data ↗

Checking the site connection security

Honor Fang Fei: Robot Phone is only first-generation exploration, future phones will move from"Container for the application"Heading towards"The stage for agents" ↗

Fang Fei said that Honor has always believed that future phones will move from 'containers of apps' to 'stages of agents'. Honor Robot Phone is only the first generation of exploration, and will continue to turn imagination into products that users can truly perceive, use and trust.

Four bloggers sent offers to Qianwen, handing repetitive chores to it ↗

Over 10 models in 70 days: CNBC says US restrictions instead catalyze Chinese AI innovation, accelerating entry into global enterprise business systems ↗

The Seoul Economic Daily published a blog post today (August 14) reporting that since June this year, nine Chinese AI companies have launched 10 AI models, which are gradually approaching top US closed-source models in key areas such as reasoning, programming, and agents.

Z.ai releases GLM-5.3: No retraining of base model, significant improvement in complex programming and long-horizon task capabilities ↗

Z.ai just released GLM-5.3. GLM-5.3 runs on the same 743B base model as GLM-5.2. Every reported gain comes from scaled post-training: more task environments, more environment types, longer training. The results land in two places. Coding jumps most on the longest-horizon benchmarks, with Terminal-Bench 3.0 moving from 4.6 to 28.3. Cybersecurity moved further than Z.ai says it expected, with CyberGym reaching 84.5%. Weights are not public yet.

How organizations use artificial intelligence: evidence from ChatGPT ↗

Xinqing Technology automotive-grade 7-nanometer AI acceleration chip"Tiangong 100"Mass production; hardware costs drop to the thousand-yuan level ↗

The 'Tiangong 100', a 7-nanometer automotive-grade AI acceleration chip independently developed by Xinqing Technology, has entered full mass production. It supports plug-in card upgrades without modifying original vehicle hardware, is compatible with multiple car brands, and provides a domestic computing base for on-device large models.

OpenAI Chief Revenue Officer Dennis departs; Wiz President Dari Raič takes over ↗

Earlier this week, long-serving executive Brad Lightcap announced he would leave to pursue a new project; Figi Simo, who was responsible for most of OpenAI's core operations, stepped down last month after sick leave; former product head Kevin Weil had already left OpenAI in April.

MiniMax-Music3 music model released: AI generates songs up to 5 minutes long ↗

AI company MiniMax launched its music generation model MiniMax-Music3 today (August 14), capable of generating complete songs up to 5 minutes long based on lyrics and musical descriptions.