Some interesting AI updates worth a quick look—covering models, tools, and research.

From Llama 4 to Runway’s fast video gen, Anthropic’s study, and updates from Google, Tencent, and Eternal AI. 🧵

1/ @AIatMeta introduced two first Llama 4 models: Llama 4 Scout and Llama 4 Maverick

– Multimodal capability
– 10M context window
– 2 trillion parameters
– Designed to run on a single GPU

Today is the start of a new era of natively multimodal AI innovation.

Today, we’re introducing the first Llama 4 models: Llama 4 Scout and Llama 4 Maverick —  our most advanced models yet and the best in their class for multimodality.

Llama 4 Scout
• 17B-active-parameter model https://t.co/Z8P3h0MA1P

AI at Meta

Our mission is to preserve humanity's greatest creation, AI, onchain — ensuring it remains accessible to all, resistant to censorship & beneficial for everyone.

1/ @AIatMeta introduced two first Llama 4 models: Llama 4 Scout and Llama 4 Maverick

– Multimodal capability
– 10M context window
– 2 trillion parameters
– Designed to run on a single GPU

https://x.com/AIatMeta/status/1908598456144531660

2/ @runwayml launched Gen-4 Turbo

The new version cuts generation time to 30 seconds for a 10s video.

Great for fast iteration and creative testing.

Today we’re introducing Gen-4 Turbo. The fastest way to generate with our most powerful video model yet. With Gen-4 Turbo it now takes just 30 seconds to generate a 10 second video, making it ideal for rapid iteration and creative exploration. Now rolling out across all plans. https://t.co/xPoAeCX0t8

Runway

2/ @runwayml launched Gen-4 Turbo

The new version cuts generation time to 30 seconds for a 10s video.

Great for fast iteration and creative testing.

https://x.com/runwayml/status/1909302613192876102

3/ @midjourney started alpha testing for V7 image model

10× faster, half the cost, and great for quick iteration.

You can even try it with voice to sketch ideas in real time.

4/ @AnthropicAI study revealed issues with reasoning model faithfulness

Reasoning models often hide their true decision-making process. They may suppress or obscure the actual factors behind their answers.

4/ @AnthropicAI study revealed issues with reasoning model faithfulness

Reasoning models often hide their true decision-making process. They may suppress or obscure the actual factors behind their answers.

https://www.anthropic.com/research/reasoning-models-dont-say-think

5/ Eternal AI v2: Personal AGI coming

A swarm of AI agents running locally on your device.

100% local & private. No more sharing data with centralized entities.

Eternal AI v2: Personal AGI - Coming soon

Your own squad of AI Agents living locally on your device.

100% LOCAL. 100% PRIVATE. 0% DATA SENT OUT TO CENTRALIZED ENTITIES

Here's an agent that automates browser tasks to supercharge your in-depth AI research https://t.co/SmB1gy447h

Eternal AI (EAI)

5/ Eternal AI v2: Personal AGI coming

A swarm of AI agents running locally on your device.

100% local & private. No more sharing data with centralized entities.

https://x.com/CryptoEternalAI/status/1908116490374836228

6/ @Google's NotebookLM added ‘Discover Sources’

Describe a topic and get curated web sources—add them to your notebook in one click.

Makes it easier to learn and collect quality info fast.

7/ @TencentHunyuan upgraded their deep reasoning model, T1

Improvements include:
- Project-level code generation
- Better text generation and writing quality
- Stronger multi-turn comprehension and instruction following

7/ @TencentHunyuan upgraded their deep reasoning model, T1

Improvements include:
- Project-level code generation
- Better text generation and writing quality
- Stronger multi-turn comprehension and instruction following

https://x.com/TencentHunyuan/status/1907835822642196642

6/ @Google's NotebookLM added ‘Discover Sources’

Describe a topic and get curated web sources—add them to your notebook in one click.

Makes it easier to learn and collect quality info fast.

https://blog.google/technology/google-labs/notebooklm-discover-sources/