{"slug": "ai-agents-weekly-kimi-k3-thinking-machines-inkling-agentic-misalignment-space-2", "title": "🤖 AI Agents Weekly: Kimi K3, Thinking Machines Inkling, Agentic Misalignment, Perplexity SPACE, Sunday ACT-2, GPT-Red, and More", "summary": "Moonshot AI open-sourced Kimi K3, a 2.8-trillion-parameter mixture-of-experts model with a 1-million-token context window and native vision, achieving state-of-the-art results on agentic benchmarks like Terminal-Bench 2.1 and DeepSWE. The model introduces Kimi Delta Attention for faster decoding and Attention Residuals for improved training efficiency, with weights to be released on July 27.", "body_md": "# 🤖 AI Agents Weekly: Kimi K3, Thinking Machines Inkling, Agentic Misalignment, Perplexity SPACE, Sunday ACT-2, GPT-Red, and More\n\n### Kimi K3, Thinking Machines Inkling, Agentic Misalignment, Perplexity SPACE, Sunday ACT-2, GPT-Red, and More\n\nIn today's issue:\n\nMoonshot open-sources Kimi K3\n\nThinking Machines ships Inkling weights\n\nAnthropic finds new agent misalignment\n\nPerplexity opens SPACE agent runtime\n\nSunday Robotics unveils ACT-2\n\nOpenAI launches GPT-Red red teamer\n\nMeituan open-sources LongCat-2.0\n\nschema harness saturates ARC-AGI-3\n\nAnd all the top AI dev news, papers, and tools.\n\n## Top Stories\n\n### Moonshot Open-Sources Kimi K3\n\nMoonshot AI introduced Kimi K3, billed as the first open 3-trillion-class model, with weights due July 27.\n\n**Scale and architecture:** A 2.8T-parameter Stable LatentMoE activating 16 of 896 experts, with a 1M-token context window and native vision in a single model.**New attention design:** Kimi Delta Attention delivers up to 6.3x faster decoding in million-token contexts, and Attention Residuals add roughly 25% higher training efficiency at under 2% extra compute.**Agentic strength:** Leads Terminal-Bench 2.1 and scores 67.3 on DeepSWE, with demos of autonomous kernel optimization, compiler development, and vision-in-the-loop coding.**Availability:** Live via Kimi.com, Kimi Code, and API from about 0.30 dollars to 15 dollars per million tokens, with open weights following on July 27.", "url": "https://wpnews.pro/news/ai-agents-weekly-kimi-k3-thinking-machines-inkling-agentic-misalignment-space-2", "canonical_source": "https://nlp.elvissaravia.com/p/ai-agents-weekly-kimi-k3-thinking", "published_at": "2026-07-18 15:00:48+00:00", "updated_at": "2026-07-18 15:26:43.014468+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "ai-research"], "entities": ["Moonshot AI", "Kimi K3", "Terminal-Bench", "DeepSWE"], "alternates": {"html": "https://wpnews.pro/news/ai-agents-weekly-kimi-k3-thinking-machines-inkling-agentic-misalignment-space-2", "markdown": "https://wpnews.pro/news/ai-agents-weekly-kimi-k3-thinking-machines-inkling-agentic-misalignment-space-2.md", "text": "https://wpnews.pro/news/ai-agents-weekly-kimi-k3-thinking-machines-inkling-agentic-misalignment-space-2.txt", "jsonld": "https://wpnews.pro/news/ai-agents-weekly-kimi-k3-thinking-machines-inkling-agentic-misalignment-space-2.jsonld"}}