According to AI at Meta on X, Meta released Muse Spark Open, a US open-weight model for coding and agentic work with 816B total and 42B active parameters, a 1M-token context window, and native multimodal support across text, image, video, and audio.

According to Google DeepMind on X, Gemini 4 Argon is now globally available, bringing frontier capabilities for coding, enterprise workflows, and cyber defense.

According to TheRundownAI, Reflection’s Beam debuts as a 501B-parameter open model with 23B active, optimized for coding and agentic tasks, with weights this month.

According to GoogleDeepMind, Gemini 4 Argon targets coding, enterprise knowledge work, and cybersecurity, rolling out to testers via the Fairwind Program.

According to OpenAI... GPT-6 Sol and Luna offer faster, cheaper models with 50% lower API prices versus GPT-5.6, improving coding, factuality, and tooling.

According to AIatMeta, Muse Spark 1.3 improves agentic and coding tasks, cuts tool calls by ~20% and tokens by ~25% versus 1.2.

According to DemisHassabis, Google launched Gemini 3.8 Flash and 3.8 Flash Cyber with faster coding and cyber defense gains, per Google blog.

According to KyeGomezB, Qwen3.8-Max debuts as a positive, cowork-focused AI with strong coding utility, as reported by Alibaba_Qwen on X.

According to godofprompt, switching to GLM5.2 cut token costs 11x; according to Zai_org, it adds 1M context, coding gains, and MIT open weights.

According to sundarpichai, Gemini 3.5 Flash is live, 4x faster than frontier models and outperforms 3.1 Pro on most benchmarks, with major coding gains.

According to GoogleDeepMind, Gemini 3.5 Flash debuts as its strongest model for agents and coding, combining frontier intelligence with real‑world action.

According to Anthropic (@AnthropicAI), new research reveals that while AI tools can speed up software engineering tasks, over-reliance may hinder skill development. Anthropic conducted an experiment with software engineers and found that coding with AI assistance led to a decrease in mastery, but the effect varied based on how engineers engaged with the AI. The study suggests that asking clarifying questions to AI during tasks can help retain skills and support ongoing learning, as reported by Anthropic. This highlights a key business opportunity for companies to train employees on effective AI usage to maximize productivity without sacrificing expertise.

According to Anthropic (@AnthropicAI), an experiment with software engineers revealed that while AI accelerates work, reliance on AI coding tools can decrease skill mastery, depending on how individuals use these tools. The findings highlight that the impact of AI assistants like Claude3 on learning new technical skills depends on user habits and interaction patterns. As reported by Anthropic, this suggests businesses adopting AI for software development should balance productivity gains with strategies to maintain workforce skill growth.