Z.AI launches GLM-4.7, new SOTA open-source model for coding

ZAI has launched its Open Platform featuring the glm-4.7 language model, offering global access and API support for Python and Java. Glm-4.7 improves reasoning, supports 4096 tokens, and shows strong performance on the HLE benchmark.

🗞 #ai

Perplexity launched its "2025 Year in Review" widget, where users can select from a set of predefined prompts to recap their activities during 2025.

Blast from the past! 👀

Gemini got a snowman ⛄️

Exclusive: Google tests 30-minute audio Lectures on NotebookLM

Google is testing a new “Lecture” mode in NotebookLM's Audio Overviews, offering 30-minute monologues that explain and connect uploaded sources. A British English voice is also planned, with full rollout expected in 2026.

Warning ⚠️

I am playing with "Subscribers First" content to bring some extra value to our top supporters. This article will be revealed to free users in 4-5 hours from now.

🗞 #notebooklm

MiniMax M2.1 was officially announced, scoring 72.5% on SWE-multilingual.

"A SOTA 10B-activated OSS coding & agent model, scoring 72.5% on SWE-multilingual and 88.6% on our newly open-sourced VIBE-bench"

NotebookLM is teasing new voices with British English narration that are planned to arrive in 2026.

These voices are available for listening in the "Archive 1945" featured notebook, created in partnership with The Economist.

Stay tuned for the TC notebook today 👀

Exclusive: Google tests 30-minute audio Lectures on NotebookLM

Google’s NotebookLM is developing a new “Lecture” mode for Audio Overviews, offering AI-generated 30-minute monologues to explain source materials, with voice and language options, aimed at students and professionals reviewing complex content.

🗞 #notebooklm

BREAKING 🚨: Google might be working on a "Lecture" format for Audio Overviews on NotebookLM. Lectures would have a single host narration and could last for 30 minutes.

Exclusive Notebook with TestingCatalog Lecture sample in the post above.

Listening time 👀

BREAKING 🚨: Poetiq system with GPT-5.2 X-High as a base, broke through human baseline on ARC-AGI-2 benchmark.

From 65% to 75% in a month 🤖

Manus launches Connectors for Projects

Manus Projects getting Connectors, allowing users to integrate apps like Gmail, Notion, and GitHub into their workflows. The feature supports pre-built and custom connections, enabling precise automation.

🗞 #manus

Windsurf Wave 13 brings free SWE-1.5 and new upgrades

Windsurf's Wave 13 enables parallel Cascade agents using Git worktrees to reduce merge conflicts across branches. It adds multi-agent panes, a context window, automatable hooks, and a dedicated terminal. SWE-1.5 is now free and the default model.

🗞 #windsurf

Marry X-mass! 🎅🎅🎅

Liquid AI released LFM2-2.6B-Exp, a small, open 3B model that outperforms DeepSeek R1-0528 in certain benchmarks, such as IFBench.

I believe that after AGI, we would get a 3B open-source model with the same performance, which can run on any device.

Small intelligence 👀