ByteDance just released Seed 2.0 and it's SOTA on multimodal, math , STEM, and agentic reasoning benchmarks.

Seedance 2.0 video is also now the best Ai video model.

Just another Deepseek moment from china.

DeepSeek V4 benchmarks just leaked.

83.7% on SWE-Bench Verified.

That would make it the best coding model in the world.

Alibaba just released Qwen 3.5

It's coming ...

Top 10 Open Source AI Models (Feb 2026):

1. GLM-5 - Autonomous engineering & app building
2. MiniMax M2.5 - Coding beast (80.2% SWE-bench)
3. Kimi K2.5 - Vision, coding & human-level reasoning
4. DeepSeek V3.2 - Smart sparse attention rivals GPT
5. Qwen 3 - Efficient logic & multilingual expert
6. MiMo V2 Flash - High efficiency, low cost
7. Mistral Large 3 - Commercial-friendly & massive context
8. LongCat Flash Chat - For ultra-long conversations
9. Gemma 3 - Google's open model with emotional nuance
10. INTELLECT-3 - Fully open training stack

AI is reshaping skills unevenly – digital and information-processing face the fastest change, caring skills the slowest.

Anthropic released Sonnet 4.6.

Google just released Lyria 3, it's music generation model.

Hello, 👋

Grok 4.20 Updates.

GPT-5.3 Instant in ChatGPT is now rolling out to everyone.

GPT 5.4 is coming Sooner...

OpenAi released GPT-5.4 Thinking and GPT-5.4 Pro.

Anthropic Capabilities by catagory.