<![CDATA[Stories by Caspar - AI Engineer on Medium]]> https://medium.com/@CasparAI?source=rss-4a69b256c722------2 https://cdn-images-1.medium.com/fit/c/150/150/1*RCffI8BwrqAWQlQeT93dNQ.png Stories by Caspar - AI Engineer on Medium https://medium.com/@CasparAI?source=rss-4a69b256c722------2 Medium Sat, 14 Nov 2026 03:31:11 GMT <![CDATA[AI Job Apocalypse? No, Just Learn These 10 AI Skills Before 2027]]>

Want to save yourself? You’ll need these skills.

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/ai-job-apocalypse-no-just-learn-these-10-ai-skills-before-2027-31f89b9c9658?source=rss-4a69b256c722------2 https://medium.com/p/31f89b9c9658 Fri, 13 Nov 2026 14:14:08 GMT 2026-11-13T14:14:08.373Z <![CDATA[Jensen Huang Resigns, and the NVIDIA Memes Are Ruthless]]>

The board went from backing him to naming an interim CEO in four days, and the internet made jokes my shares didn’t need

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/jensen-huang-resigns-and-the-nvidia-memes-are-ruthless-98ac7bcb9303?source=rss-4a69b256c722------2 https://medium.com/p/98ac7bcb9303 Thu, 12 Nov 2026 17:14:40 GMT 2026-11-12T17:14:40.316Z <![CDATA[Meta Just Shipped the Best Small Model and Nobody Noticed]]>

Muse Ember matches Claude Haiku 5.5 and its way cheaper..

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/meta-just-shipped-the-best-small-model-and-nobody-noticed-8215ff90dca7?source=rss-4a69b256c722------2 https://medium.com/p/8215ff90dca7 Wed, 11 Nov 2026 21:32:03 GMT 2026-11-11T21:32:03.585Z <![CDATA[What the Hell Is Going On at NVIDIA?]]>

A reported Chinese intelligence effort, a board investigation and a $190 billion selloff.

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/what-the-hell-is-going-on-at-nvidia-4440bd37166d?source=rss-4a69b256c722------2 https://medium.com/p/4440bd37166d Tue, 10 Nov 2026 15:25:14 GMT 2026-11-10T15:25:14.206Z <![CDATA[Claude Fable 5.2 Widens Anthropic’s Lead Over OpenAI]]>

At Medium effort it beats Opus 5.5 High for about 34 cents more per task. Max tops the charts at a price I can’t justify.

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/claude-fable-5-2-widens-anthropics-lead-over-openai-1263002e9940?source=rss-4a69b256c722------2 https://medium.com/p/1263002e9940 Mon, 09 Nov 2026 23:01:21 GMT 2026-11-09T23:01:21.014Z <![CDATA[Is Training Outdated? AI Can Just Improve Its Own Weights Apparently…]]>

What if you could debug the model you already use, without retraining?

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/is-training-outdated-ai-can-just-improve-its-own-weights-apparently-95af8f941875?source=rss-4a69b256c722------2 https://medium.com/p/95af8f941875 Mon, 09 Nov 2026 19:44:07 GMT 2026-11-09T19:44:07.485Z <![CDATA[Even Engineering Is Not Safe From AI Anymore — Meet SkyFall AI]]>

This new lab claims it can solve engineering with its new world model, and the ‘trust me bro’ benchmarks seem impressive.

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/even-engineering-is-not-safe-from-ai-anymore-meet-skyfall-ai-ee929cd7af58?source=rss-4a69b256c722------2 https://medium.com/p/ee929cd7af58 Sun, 08 Nov 2026 10:59:39 GMT 2026-11-08T10:59:39.692Z <![CDATA[This New Local Open-Weight Model Rivals Giants 10× Its Size]]>

Ricursive’s Ripple has 65B params and competes with Meta’s Open weight model

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/this-new-local-open-weight-model-rivals-giants-10-its-size-197a58614eaa?source=rss-4a69b256c722------2 https://medium.com/p/197a58614eaa Sat, 07 Nov 2026 17:13:43 GMT 2026-11-07T17:13:43.885Z <![CDATA[DeepSeek Is on a Generational Run — And Nobody Seems To Notice]]>

Besides cost efficient models, Deepseek is becoming an Applied AI powerhouse

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/deepseek-is-on-a-generational-run-and-nobody-seems-to-notice-85182893ae3c?source=rss-4a69b256c722------2 https://medium.com/p/85182893ae3c Fri, 06 Nov 2026 16:22:00 GMT 2026-11-06T16:22:00.927Z <![CDATA[NVIDIA’s Dory Is A Local LLM That Loops]]>

A 4B open-weight model can rerun its middle layers for higher scores, outperforming larger rivals

Continue reading on Inside AI »

]]> https://medium.com/the-ai-brief/nvidias-dory-is-a-local-llm-that-loops-a5d8c2db768b?source=rss-4a69b256c722------2 https://medium.com/p/a5d8c2db768b Thu, 05 Nov 2026 18:56:28 GMT 2026-11-05T18:56:28.906Z