Qwen 3 Max Thinking: Flagship reasoning model with advanced capabilities 🤖
Key details:
- Matches or beats GPT-5.2-Thinking, Claude Opus 4.5, Gemini 3 Pro on several benchmarks.
- Strong in reasoning, agentic search, tool use, and coding.
Run & access: Closed-source. Cloud-only via subscription or free trial.
@CoreAti #qwen #alibaba
Key details:
- Matches or beats GPT-5.2-Thinking, Claude Opus 4.5, Gemini 3 Pro on several benchmarks.
- Strong in reasoning, agentic search, tool use, and coding.
Run & access: Closed-source. Cloud-only via subscription or free trial.
@CoreAti #qwen #alibaba
❤1👎1
MOVA: New open-source video generator with sound 👀
- This new video generator includes native sound.
- Offering 360p and 720p resolution videos.
- It is a mixture of experts model with 32 billion parameters.
Run & access: Open-source. ~32B parameters, requires ~77 GB VRAM, not practical for consumer GPUs.
- This new video generator includes native sound.
- Offering 360p and 720p resolution videos.
- It is a mixture of experts model with 32 billion parameters.
Run & access: Open-source. ~32B parameters, requires ~77 GB VRAM, not practical for consumer GPUs.
@CoreAti #OpenSource #mova #videoAi
Lucy 2: Real-time AI video editor 🔥
Decart's Lucy 2 is a real-time video editor that allows users to modify videos with natural language prompts, supporting high-fidelity output.
Key details:
- Generates 1080p videos at 30 FPS
- Near-zero latency for real-time editing
Run & access: Closed-source. Cloud-based demo and paid access.
Decart's Lucy 2 is a real-time video editor that allows users to modify videos with natural language prompts, supporting high-fidelity output.
Key details:
- Generates 1080p videos at 30 FPS
- Near-zero latency for real-time editing
Run & access: Closed-source. Cloud-based demo and paid access.
@CoreAti #VideoEditor #videoAi #lucy2
Core Ai News
Tencent launched two new Hunyuan models: —HunyuanWorld-Voyager, an ultra-long-range world model with native 3D reconstruction and memory —Hunyuan-MT-7B and Hunyuan-MT-Chimera-7B, a joint AI translation system that outperforms rivals across 33 languages
Hunyuan Image 3.0 Instruct: Tencent’s model rivaling top image generators 🔥
Key details:
- Features chain-of-thought reasoning
- Capable of complex image manipulation and composition
- Requires significant VRAM (169 GB) making local use challenging.
https://huggingface.co/tencent/HunyuanImage-3.0-Instruct
Key details:
- Features chain-of-thought reasoning
- Capable of complex image manipulation and composition
- Requires significant VRAM (169 GB) making local use challenging.
https://huggingface.co/tencent/HunyuanImage-3.0-Instruct
@CoreAti #imageai #hunyuan #tencent
This media is not supported in your browser
VIEW IN TELEGRAM
TeleStyle: AI for consistent style transfer in images and videos 👀
TeleStyle can change artistic style of images and videos to match a reference style, either from a prompt or an image.
Key details:
- Supports various artistic styles, including anime and Victorian
- Maintains consistency across video frames
Run & access: Open-source. Runs on consumer GPUs with 8 GB VRAM.
https://tele-ai.github.io/TeleStyle/
TeleStyle can change artistic style of images and videos to match a reference style, either from a prompt or an image.
Key details:
- Supports various artistic styles, including anime and Victorian
- Maintains consistency across video frames
Run & access: Open-source. Runs on consumer GPUs with 8 GB VRAM.
https://tele-ai.github.io/TeleStyle/
@CoreAti #videoai #TeleStyle #imageai
AI This Week: Open models, faster video, stronger reasoning 🚨
Earth-2 — Open-source weather AI🖥
Qwen 3 Max — Elite reasoning model
MOVA — Video AI with sound
Lucy 2 — Live video editing
Hunyuan 3.0 — High-end image AI🇨🇳
TeleStyle — Style transfer AI
Genie 2 — Ai World Explorer🔍
MoltBook — Social Media For Ai Agents😎
Earth-2 — Open-source weather AI
Qwen 3 Max — Elite reasoning model
MOVA — Video AI with sound
Lucy 2 — Live video editing
Hunyuan 3.0 — High-end image AI
TeleStyle — Style transfer AI
Genie 2 — Ai World Explorer
MoltBook — Social Media For Ai Agents
Please open Telegram to view this post
VIEW IN TELEGRAM
🔥3❤1👏1
You don't need to buy a Mac mini, use a terminal, or worry about configuration files.
Here's what you can do:
1. Go to Emergent's homepage
2. Click on the MoltBot option
3. A prompt will automatically show up
4. Click next
That's it.
Emergent Link. Emergent Link
Please open Telegram to view this post
VIEW IN TELEGRAM
❤3🔥1🥰1
You can then connect your WhatsApp or Telegram account to start using the bot immediately.
Emergent Link | Emergent Link
This turns 2 hours of installation and configuration into 2 minutes.
But even more important:
This will help you keep things secure and away from your computer. MoltBot is fun and powerful, but I do not recommend giving it access to anything critical.
Emergent Link | Emergent Link
❤2👍1🥰1
StepFun launched Step 3.5 Flash open-source model
Key details:
• Sparse MoE architecture matches GPT-5, Claude Opus 4.5 on reasoning benchmarks.
• 256K context window, efficient Sliding Window Attention.
Key details:
• Sparse MoE architecture matches GPT-5, Claude Opus 4.5 on reasoning benchmarks.
• 256K context window, efficient Sliding Window Attention.
Run & access:
Runs on consumer-grade hardware
-Mac Studio M4 Max,
-NVIDIA DGX Spark 128GB
Local deployment via llama.cpp or cloud API at platform.stepfun.ai
@CoreAti #stepfun #opensource #ai #reasoning
The Core: NASA just revealed that its rover completed the first-ever AI-planned drive on another planet.
with Anthropic’s Claude helping map a 400m route across the Martian surface that the six-wheeled robot navigated in December.
with Anthropic’s Claude helping map a 400m route across the Martian surface that the six-wheeled robot navigated in December.
OpenAI announced that it will retire GPT-4o and several other legacy models from ChatGPT in mid-February, noting that fewer than 0.1% of users still choose them daily.
🛠️ Trending AI Tools
🤖 Claude Cowork - New plugins to bundle skills, MCPs, and tools
🔍 Project Genie - Google DeepMind's interactive world generator
📹 Grok Imagine - xAI’s upgraded video model, now available via API
🦞 OpenClaw (formerly Clawdbot) - Open personal AI assistant for chat apps
🦞 OpenClaw (formerly Clawdbot) - Open personal AI assistant for chat apps
Please open Telegram to view this post
VIEW IN TELEGRAM
This media is not supported in your browser
VIEW IN TELEGRAM
BREAKING 🚨: OpenAI says Codex built a full racing game with 7M tokens from one prompt, featuring 8 maps, 8 unique characters, items, and AI opponents.
They used it to demo the new Codex desktop app released today for Mac:
— Run multiple agents in parallel
— Skills for tasks beyond code
— Automations on a schedule
— Worktrees so agents don't conflict
They used it to demo the new Codex desktop app released today for Mac:
— Run multiple agents in parallel
— Skills for tasks beyond code
— Automations on a schedule
— Worktrees so agents don't conflict
❤4
Launching this month
- Sonnet 5
- GPT-5.3
- Gemini 3 GA
- DeepSeek v4 (/r2)
- GLM-5
- Sonnet 5
- GPT-5.3
- Gemini 3 GA
- DeepSeek v4 (/r2)
- GLM-5
❤3
GPT-5.2 and GPT-5.2-Codex got a 40% speed boost with a new inference.
That was fast 👀
That was fast 👀
Please open Telegram to view this post
VIEW IN TELEGRAM
🛠️ Trending AI Tools
📱 Codex App - OpenAI’s Mac app for managing AI agents
💬 Riverflow 2.0 - Sourceful’s new top-ranked AI image editing model
🧠 Step-3.5-Flash - StepFun’s AI with strong reasoning, agentic capabilities
🎙 Eleven v3 - ElevenLabs’ expressive voice AI, now commercially available
🧠 Step-3.5-Flash - StepFun’s AI with strong reasoning, agentic capabilities
Please open Telegram to view this post
VIEW IN TELEGRAM
🚨 NEW: Nvidia nears $20B investment in OpenAI's latest funding round, marking its single biggest bet on the ChatGPT developer, Bloomberg reports.