14th August - AI News Daily - Beyond Text: How OpenAI, Google DeepMind, and Anthropic Are Pushing AI's Multimodal Frontiers

Send us a text AI News Summaries https://s.server489.com/AI-2025-08-14 AI Tweet Summaries https://s.server489.com/XAI-2025-08-14 Frameworks & Platforms: DSPy 3.0 exited beta with MCP and audio support; TRL shipped native fine-tuning with multimodal GRPO/MPO; Oumi launched DCVLR challenge for vision-language datasets; Nomic rebranded with upcoming open-source releases; Google DeepMind updated Perch for wildlife monitoring. Industry Headlines: Leadership changes at Cohere Labs; Elon Musk threatening legal action over Apple's OpenAI promotion; Grok surpassing Google in App Store rankings; AMD's Dr. Sharon Zhou announced for PyTorch Conference. New Tools: Higgsfield launched draw-to-video workflow; new open-source agent chains LLMs with image/video generators; Mule Run released beta marketplace for AI agents; Anycoder provided free open-source coding app on Gradio; Cline positioned as focused AI engineering platform. LLM Advancements: GPT-5 unveiled with broad capability gains; Qwen-3-235b topped leaderboards; GLM-4.5 and gpt-oss-120b entered top 10; Mistral Medium 3.1 targeted coding; Gemma 3 27B excelled on consumer GPUs. Beyond Text: Genie 3 (11B) showed strong 3D reasoning; Wan 2.2 14B reduced video generation latency; LiquidAI's LFM2-VL delivered on-device vision; OpenAI's gpt-oss 120B generated full videos. Product Updates: AI Studio added GitHub integration; W&B Weave introduced unified assets view; LlamaExtract added to TypeScript SDK; Grok Imagine removed video limits; Ollama launched Turbo Mode; LangChain debuted Deep Agents UI; Perplexity rolled out Comet; Anthropic added prompt cache; Claude Code incorporated Opus 4.1; FastPlaid made indexes mutable; Gemini gained memory features. Resources: Guides for local RAG pipelines with GPT-OSS; DAIR.AI launched agent design training; specialist model recipes shared; Weaviate Podcast on vector search. Applications: SkySQL achieved hallucination-free SQL generation; locodiff curve experiments pushed generative limits. Industry Discussions: Kaggle's Game Arena showed skill transfer; debates on AGI timeline (majority expect before 2030); Stanford analysis criticized YC AI startups; research on LLM energy costs; concerns about paid promotions; methodological critiques of evaluation metrics. Major Industry News: GPT-5 launch faced backlash over inconsistency; OpenAI launched $500K red-teaming challenge; OpenAI reportedly backing Merge Labs BCI startup; Anthropic offered $1 Claude subscription to US federal government; APT28 deployed LLM-powered malware; Disney/Universal sued Midjourney; Google rolled out Gemini Personal Context; medical study showed AI dependency risks; Austria's AI tax enforcement added €354M; Arm previewed 2026 mobile GPUs. Support the show 🌍 INAI • The Open AI Hub The Intelligence Atlas → the world’s most comprehensive, open hub of AI knowledge. 2 Million+ tools, models, agents, tutorials & daily news —free for all, updated every day. https://github.com/inai-sandy/inAI-wiki