AI Engineering Digest https://ai-eng-digest.vercel.app/ Sourced briefing for engineers, from As The Geek Learns. Primary sources only; vendor evals flagged. en-us Mon, 24 Aug 2026 12:00:00 +0000 Claude computer use leaves beta as llama.cpp wires GLM MTP and SenseNova lands on MLX https://ai-eng-digest.vercel.app/archive/2026-08-24.html https://ai-eng-digest.vercel.app/archive/2026-08-24.html Mon, 24 Aug 2026 12:00:00 +0000 Sourced briefing for engineers. Window ~2026-07-28 to 2026-08-24. Primary sources only. 1. Enterprise — Claude computer use / browser use / Files/Skills GA; Google Antigravity in Gemini Enterprise; Amazon Bedrock AgentCore temporal policies; Databricks Unity AI Gateway GA; Pinecone Nexus GA (vendor evals). 2. Local — llama.cpp b10603 GLM-4.5-Air MTP; Ollama 0.33.0-rc2; mlx-lm Muse Glimmer PR #1710 still open. 3. MLX / Apple Silicon — SenseNova-U1.5-8B-MoT MLX-Swift weights (today); mlx-vlm 0.6.15. 4. Frontier — DeepSeek-V4-Flash-Vision-Exp; GPT-5.6 Sol API/credit cut over 20% (no new dollar rates in the post); Ultrafast Sol preview; Grok Voice Think Fast 2.0. 5. Suggested projects — SenseNova-U1.5 8-bit T2I on MLX-Swift; flag GitHub 3.2s/14.8GB vs card 7.4s/22.9GB. 6. Open source tools — Ray 2.58.0; Agent Lightning 1.0; SGLang 0.5.18; vLLM 0.27.0; Switchyard 0.2.0. 7. Open source models — Ornith-1.5; Nemotron 3.5 Lightning; North-Micro-Vision-Instruct. 8. Other — NVIDIA + SB Energy PORTS-Pike; OpenAI tenant. Glimmer ships on-device as frontier APIs drop price and EU Article 50 goes live https://ai-eng-digest.vercel.app/archive/2026-08-22.html https://ai-eng-digest.vercel.app/archive/2026-08-22.html Sat, 22 Aug 2026 14:00:00 +0000 Sourced briefing for engineers. Window ~2026-07-25 to 2026-08-22. Primary sources only. 1. Enterprise AI engineering — Anthropic Inference Hooks; Google Agent Platform evals GA; LangSmith Tuned Evaluators; Anthropic computer use / Skills / Files GA; OWASP GenAI LLM Top 10 2026; Snowflake Cortex dynamic routing (preview); OpenAI Presence (just outside window). 2. Local LLMs — Muse Glimmer 30B; Ollama 0.32.7–0.32.15; llama.cpp v0.2.0; llamafile 0.10.5; LM Studio Bionic. 3. MLX / Apple Silicon — MLX 0.32.1; Qwen3.8-27B-4bit; Muse-Glimmer-30B-4bit; mlx-swift-lm PR 523. 4. Frontier models — Grok 4.6; Gemini 3.7 Flash; DeepSeek-V4-Pro + Flash vision; OpenAI Ultrafast Sol; Claude Opus 5 (one day before window). 5. Suggested projects — five local Apple Silicon builds around Glimmer and Qwen3.8. 6. Open source tools — vLLM 0.27.0; SGLang 0.5.17; Transformers 5.15.0; LlamaIndex v0.14.24; Open WebUI v0.11.0; LangChain core 1.6.0. 7. Open source models — Kimi K3; DeepSeek-V4-Flash-0731; DeepSeek-V4-Pro-0813; Qwen3.8-2.4T-A95B; Qwen3.8-27B; Muse Glimmer 30B; GLM-5.3 not shipped. 8. Other — EU AI Act Article 50 from 2 Aug 2026; arXiv 2608.11693v2 on INT8 / Blackwell Ultra.