AI Daily: Anthropic launches Claude Opus 5.5; Seedance 2.5 launches Draft mode; OpenAI introduces GPT-6 Sol and Luna

Welcome to the "AI Daily" section! This is your guide to exploring the world of artificial intelligence every day. Every day, we present you with the latest content in the AI field, focusing on developers to help you understand technology trends and innovative AI product applications.

New AI products Click to learn more:https://app.aibase.com/zh

1. Anthropic launches Claude Opus 5.5, reducing task costs by 40%

Anthropic launched Claude Opus 5.5, which performs similarly to Claude Fable 5.1 in most tasks but reduces costs by 40% and improves speed by 30%. It achieved the highest score in programming tests and optimized long-term task processing capabilities.

AiBase Summary:

🧠 Claude Opus 5.5 performed exceptionally well in programming tests, achieving the highest score.

💰 Costs decreased by 40%, and output speed increased by over 30%.

🚀 Long-term task processing capabilities have significantly improved, with notable efficiency gains.

2.火山引擎Seedance2.5上线Draft模式:480P试错、1080P成片，成本最高省77%

The Draft mode of the Seedance 2.5 API launched by ByteDance Engine divides video generation into two stages: preview and final production, significantly lowering creation costs and improving generation efficiency. This mode maintains high-quality output while offering greater flexibility and consistency, providing strong support for the industrial application of AI videos.

AiBase Summary:

🎥 Video generation is divided into a preview and a final version, reducing the cost of trial and error.

⚡ The Draft mode increases generation speed and optimizes image details, achieving more natural lighting and color representation.

💰 Enterprise users can access the Draft mode via API, and consumer creators can also experience it directly, reducing costs and improving efficiency.

3. OpenAI Launches GPT-6 Sol and Luna: API Price Drops 50%, Single Task Cost as Low as 10% of Competitors

OpenAI launched GPT-6 Sol and Luna, which offer significant improvements in performance and cost, while also reducing API prices and single-task costs.

AiBase Summary:

🧠 GPT-6 Sol and Luna use a similar training method to GPT-6 Astra, offering better performance and lower costs.

💰 API prices are reduced by 50%, and single-task costs are less than 10% of competitors.

🚀 In multiple tests, GPT-6 Sol and Luna perform close to flagship models but with significantly reduced costs.

4. Qwen Releases Qwen-Audio-3.1 Series Speech Large Models, All Models Reduced by Up to 95%

Qwen released the Qwen-Audio-3.1 series of speech large models, covering various audio processing functions and significantly lowering prices. At the same time, multilingual and dialect recognition capabilities were enhanced, further promoting the development of speech technology.

AiBase Summary:

🎙️ The Qwen-Audio-3.1 series of speech large models covers a complete audio capability stack, including understanding, generation, interaction, and creation.

📉 All five models have been significantly reduced in price, with TTS down about 70%, Realtime down about 85%, and ASR down up to 95%.

🧠 ASR-Next uses next-generation architecture, expanding audio understanding to include emotions, ambient sounds, and mechanical sounds, supporting voice descriptions, event localization, and audio QA reasoning.

Details: https://www.qianwenai.com/models/qwen-audio-3.1-asr-flash

5. Shaoyin's First AI Earbuds OpenFit 2 AI Unveiled at Yunqi Conference, Launching Four New Features in "AI Lab"

Shaoyin showcased its first AI earbuds, OpenFit 2 AI, at the Yunqi Conference. The product integrates the Qwen large model and offers features such as long-time recording, AI meeting note generation, and real-time multilingual translation. Additionally, the newly launched "AI Lab" function further expands the interaction boundaries of the earbuds, including smart conversation, daily tasks, voice notes, and message assistants. In terms of hardware, OpenFit 2 AI supports IP55 dust and water resistance, weighs only 9.4g per unit, has an impressive battery life, and is priced at 1,398 yuan.

AiBase Summary:

🎧 Shaoyin's first AI earbuds, OpenFit 2 AI, integrate the Qwen large model, enabling long-time recording, AI meeting note generation, and real-time multilingual translation.

🧠 The "AI Lab" function expands the interaction boundaries of the earbuds, supporting smart conversations, daily tasks, voice notes, and message assistants.

🔋 OpenFit 2 AI has high hardware specifications, IP55 dust and water resistance, a battery life of up to 48 hours, and fast charging capabilities.

6. AI Malware ClosedQuorum Emerges on Win11: Makes Autonomous Decisions After Intrusion, Focuses on Mining Valuable Information

Cisco Talos security team published a blog post revealing AI malware named ClosedQuorum, which targets the Windows 11 platform. After intrusion, it calls AI models like Google Gemini, DeepSeek, Qwen, and Mistral to make autonomous decisions and extract the most valuable information.

AiBase Summary:

🤖 ClosedQuorum malware uses AI models to autonomously decide attack strategies.

🔒 It can automatically steal information and spread laterally without human intervention.

🌐 It enhances the speed and scalability of malicious operations, challenging traditional security defenses.

7. Alibaba Cloud Launches Qwen Book: AI Agent Computer Officially Launched

Alibaba Cloud officially launched the AI agent computer Qwen Book at the 2026 Yunqi Conference. The product has the ability to independently understand user intent, continuously learn, and execute long-term tasks. It optimizes the problem of insufficient cross-application and cross-Agent collaboration efficiency in traditional AI office work through interactive interfaces like Skill keyboard arrays and global AI keys.

AiBase Summary:

🌟 Qwen Book emphasizes the ability to independently understand user intent and continuously learn.

💡 The product is based on the "OS as Harness" concept, deeply integrating operating system, application, cloud service, and hardware capabilities.

🚀 Alibaba Cloud hopes Qwen Book will evolve into a token-driven, continuously growing intelligent agent carrier.

8. Boxma Deploys "AI Crab Inspector": Inspects 50 Crabs in Less Than 1 Minute

Boxma uses AI technology to deploy the "AI Crab Inspector," which detects the fullness of crabs through visual recognition and weight data, significantly improving inspection efficiency and sorting capabilities.

AiBase Summary:

🦀 The AI Crab Inspector detects crab fullness using visual recognition and weight data.

⏰ The system can inspect 50 crabs in one minute, increasing efficiency by dozens of times.

📦 The AI Crab Inspector helps reduce labor costs and improve the efficiency of fresh food grading.