Kol Tregaskes
@koltregaskes.bsky.social
AI Art Creator 🎨 | AI Video Producer 🎬🤖 | AI Music Composer 🎵 | Tech & Science News Curator🔬 | Movie & Sci-fi Enthusiast 🍿 | Geek at Heart 🤓 | #AI #Tech
This "Prompt is too long" bug in the Claude Code desktop is annoying. Apparently it's a known issue, but I've lost a lot of sessions to it!
February 12, 2026 at 3:26 AM
This "Prompt is too long" bug in the Claude Code desktop is annoying. Apparently it's a known issue, but I've lost a lot of sessions to it!
Zai launches GLM-5, scaling to 744B parameters and 28.5T tokens for advanced agentic tasks.
- Builds on GLM-4.5 by increasing parameters from 355B to 744B and data from 23T to 28.5T tokens.
- Integrates DeepSeek Sparse Attention to cut deployment costs while maintaining long-context abilities.
- Builds on GLM-4.5 by increasing parameters from 355B to 744B and data from 23T to 28.5T tokens.
- Integrates DeepSeek Sparse Attention to cut deployment costs while maintaining long-context abilities.
February 11, 2026 at 9:26 PM
Zai launches GLM-5, scaling to 744B parameters and 28.5T tokens for advanced agentic tasks.
- Builds on GLM-4.5 by increasing parameters from 355B to 744B and data from 23T to 28.5T tokens.
- Integrates DeepSeek Sparse Attention to cut deployment costs while maintaining long-context abilities.
- Builds on GLM-4.5 by increasing parameters from 355B to 744B and data from 23T to 28.5T tokens.
- Integrates DeepSeek Sparse Attention to cut deployment costs while maintaining long-context abilities.
February 11, 2026 at 2:30 PM
February 11, 2026 at 2:11 PM
New DeepSeek model released. Model available on mobile and free to use for now. Cut off is May 2025, so a new base, 1M tokens claimed. Doesn't support image input yet. Possibly an early checkpoint. 🤔
x.com/i/status/20...
x.com/i/status/20...
February 11, 2026 at 1:58 PM
New DeepSeek model released. Model available on mobile and free to use for now. Cut off is May 2025, so a new base, 1M tokens claimed. Doesn't support image input yet. Possibly an early checkpoint. 🤔
x.com/i/status/20...
x.com/i/status/20...
Sam Altman's Town Hall last month predicts 100x cheaper intelligence by 2027 and a massive shift in software engineering roles.
- Jevons Paradox in coding: Efficiency won't kill engineering but will explode software demand; the role shifts from syntax to "directing" computers.
- Jevons Paradox in coding: Efficiency won't kill engineering but will explode software demand; the role shifts from syntax to "directing" computers.
February 11, 2026 at 12:31 PM
Sam Altman's Town Hall last month predicts 100x cheaper intelligence by 2027 and a massive shift in software engineering roles.
- Jevons Paradox in coding: Efficiency won't kill engineering but will explode software demand; the role shifts from syntax to "directing" computers.
- Jevons Paradox in coding: Efficiency won't kill engineering but will explode software demand; the role shifts from syntax to "directing" computers.
Dwarkesh argues AGI is not imminent, citing the lack of "continual learning" capabilities as a critical bottleneck preventing models from mastering context-specific tasks.
February 11, 2026 at 9:31 AM
Dwarkesh argues AGI is not imminent, citing the lack of "continual learning" capabilities as a critical bottleneck preventing models from mastering context-specific tasks.
Google Gemini achieves highest QoQ website visit growth at +72.08% in Q3-Q4 2025, outpacing others like Grok's +108.79% initial surge and Claude's +39.54% mid-year rise.
February 11, 2026 at 3:30 AM
Google Gemini achieves highest QoQ website visit growth at +72.08% in Q3-Q4 2025, outpacing others like Grok's +108.79% initial surge and Claude's +39.54% mid-year rise.
AI tools promised workload reduction but instead intensify work, leading to burnout among enthusiastic adopters.
- UC Berkeley study in a 200-person tech firm over eight months showed workers voluntarily expanded tasks and hours using AI, without mandates.
- UC Berkeley study in a 200-person tech firm over eight months showed workers voluntarily expanded tasks and hours using AI, without mandates.
February 11, 2026 at 12:30 AM
AI tools promised workload reduction but instead intensify work, leading to burnout among enthusiastic adopters.
- UC Berkeley study in a 200-person tech firm over eight months showed workers voluntarily expanded tasks and hours using AI, without mandates.
- UC Berkeley study in a 200-person tech firm over eight months showed workers voluntarily expanded tasks and hours using AI, without mandates.
This is way better, the ability to update research mid-point, select apps to use for researching and select specific sites to research on.
New Deep Research in ChatGPT.
x.com/OpenAI/stat...
New Deep Research in ChatGPT.
x.com/OpenAI/stat...
February 10, 2026 at 11:02 PM
This is way better, the ability to update research mid-point, select apps to use for researching and select specific sites to research on.
New Deep Research in ChatGPT.
x.com/OpenAI/stat...
New Deep Research in ChatGPT.
x.com/OpenAI/stat...
It took more than 10 days but Claude Cowork is now out for Windows! A user friendly version of Claude Code.
I had to download and install rather than update inside the app to get it.
x.com/claudeai/st...
I had to download and install rather than update inside the app to get it.
x.com/claudeai/st...
February 10, 2026 at 10:42 PM
It took more than 10 days but Claude Cowork is now out for Windows! A user friendly version of Claude Code.
I had to download and install rather than update inside the app to get it.
x.com/claudeai/st...
I had to download and install rather than update inside the app to get it.
x.com/claudeai/st...
The ads are here!
x.com/OpenAI/stat...
x.com/OpenAI/stat...
February 10, 2026 at 9:31 PM
The ads are here!
x.com/OpenAI/stat...
x.com/OpenAI/stat...
Skills are finally coming to ChatGPT.
x.com/mweinbach/s...
x.com/mweinbach/s...
February 10, 2026 at 6:35 PM
Skills are finally coming to ChatGPT.
x.com/mweinbach/s...
x.com/mweinbach/s...
Anyone for Garlic? 🧄🧄🧄
February 10, 2026 at 3:28 PM
Anyone for Garlic? 🧄🧄🧄
GAIR researchers introduce Context Engineering 2.0 as the next step beyond prompt engineering.
- Context engineering reduces entropy by translating human intents into machine formats.
- Builds on Karl Marx's idea that entities are defined by contexts, extending to AI-human interactions.
- Context engineering reduces entropy by translating human intents into machine formats.
- Builds on Karl Marx's idea that entities are defined by contexts, extending to AI-human interactions.
February 10, 2026 at 12:33 PM
GAIR researchers introduce Context Engineering 2.0 as the next step beyond prompt engineering.
- Context engineering reduces entropy by translating human intents into machine formats.
- Builds on Karl Marx's idea that entities are defined by contexts, extending to AI-human interactions.
- Context engineering reduces entropy by translating human intents into machine formats.
- Builds on Karl Marx's idea that entities are defined by contexts, extending to AI-human interactions.
AI enables 5-10% productivity gains in film and TV pre-production, per McKinsey analysis.
- Development and pre-production lead AI adoption, aiding script breakdowns, storyboarding, and visual pitches.
- Physical production lags due to tool limits, IP concerns, and creative integrity needs.
- Development and pre-production lead AI adoption, aiding script breakdowns, storyboarding, and visual pitches.
- Physical production lags due to tool limits, IP concerns, and creative integrity needs.
February 10, 2026 at 9:33 AM
AI enables 5-10% productivity gains in film and TV pre-production, per McKinsey analysis.
- Development and pre-production lead AI adoption, aiding script breakdowns, storyboarding, and visual pitches.
- Physical production lags due to tool limits, IP concerns, and creative integrity needs.
- Development and pre-production lead AI adoption, aiding script breakdowns, storyboarding, and visual pitches.
- Physical production lags due to tool limits, IP concerns, and creative integrity needs.
Google Antigravity and Google AI Studio to merge or integrate in some way next week.
x.com/OfficialLog...
x.com/OfficialLog...
February 10, 2026 at 7:47 AM
Google Antigravity and Google AI Studio to merge or integrate in some way next week.
x.com/OfficialLog...
x.com/OfficialLog...
GPT-5.3 spotted! Is this Garlic? Could it be released later today?
x.com/whylifeis4/...
x.com/whylifeis4/...
February 10, 2026 at 3:28 AM
GPT-5.3 spotted! Is this Garlic? Could it be released later today?
x.com/whylifeis4/...
x.com/whylifeis4/...
BabyAGI returns with version 3, a minimal autonomous AI assistant integrating advanced features for communication, task handling, and learning.
- Supports SMS and email for multi-channel interaction with shared knowledge across contexts.
- Supports SMS and email for multi-channel interaction with shared knowledge across contexts.
February 10, 2026 at 12:30 AM
BabyAGI returns with version 3, a minimal autonomous AI assistant integrating advanced features for communication, task handling, and learning.
- Supports SMS and email for multi-channel interaction with shared knowledge across contexts.
- Supports SMS and email for multi-channel interaction with shared knowledge across contexts.
Higgsfields X account has been suspended for breaching X's rules!
x.com/nickfloats/...
x.com/nickfloats/...
February 9, 2026 at 8:07 PM
Higgsfields X account has been suspended for breaching X's rules!
x.com/nickfloats/...
x.com/nickfloats/...
GLM 5 model mentioned on GitHub, must be imminent. GLM 4.7 was a very well received model.
x.com/i/status/20...
x.com/i/status/20...
February 9, 2026 at 3:31 PM
GLM 5 model mentioned on GitHub, must be imminent. GLM 4.7 was a very well received model.
x.com/i/status/20...
x.com/i/status/20...
OpenAI plans updated Chat model launch in ChatGPT this week amid growth resurgence.
- OpenAI aiming to launch updated Chat model in ChatGPT this week, following December code red to improve the chatbot.
- Ads testing starts Monday - clearly labelled at response bottoms, not affecting answers.
- OpenAI aiming to launch updated Chat model in ChatGPT this week, following December code red to improve the chatbot.
- Ads testing starts Monday - clearly labelled at response bottoms, not affecting answers.
February 9, 2026 at 2:01 PM
OpenAI plans updated Chat model launch in ChatGPT this week amid growth resurgence.
- OpenAI aiming to launch updated Chat model in ChatGPT this week, following December code red to improve the chatbot.
- Ads testing starts Monday - clearly labelled at response bottoms, not affecting answers.
- OpenAI aiming to launch updated Chat model in ChatGPT this week, following December code red to improve the chatbot.
- Ads testing starts Monday - clearly labelled at response bottoms, not affecting answers.
Is it Seedance 2.0 day today?
February 9, 2026 at 12:34 PM
Is it Seedance 2.0 day today?
Software development sees rapid AI-driven changes.
- OpenAI notes significant improvements in tools like Codex since December, enabling engineers to delegate full code writing, debugging, and operations tasks.
- OpenAI notes significant improvements in tools like Codex since December, enabling engineers to delegate full code writing, debugging, and operations tasks.
February 9, 2026 at 9:29 AM
Software development sees rapid AI-driven changes.
- OpenAI notes significant improvements in tools like Codex since December, enabling engineers to delegate full code writing, debugging, and operations tasks.
- OpenAI notes significant improvements in tools like Codex since December, enabling engineers to delegate full code writing, debugging, and operations tasks.
Qwen3-Coder-Next released with 3B active parameters.
- Built on 80B total MoE architecture for coding agents and local dev.
- Trained on 800K verifiable tasks with executable environments.
- Scores over 70% on SWE-Bench Verified using SWE-Agent scaffold.
- Built on 80B total MoE architecture for coding agents and local dev.
- Trained on 800K verifiable tasks with executable environments.
- Scores over 70% on SWE-Bench Verified using SWE-Agent scaffold.
February 9, 2026 at 12:26 AM
Qwen3-Coder-Next released with 3B active parameters.
- Built on 80B total MoE architecture for coding agents and local dev.
- Trained on 800K verifiable tasks with executable environments.
- Scores over 70% on SWE-Bench Verified using SWE-Agent scaffold.
- Built on 80B total MoE architecture for coding agents and local dev.
- Trained on 800K verifiable tasks with executable environments.
- Scores over 70% on SWE-Bench Verified using SWE-Agent scaffold.