Image created with gemini-3.1-flash-image-preview with claude-sonnet-4-5. Image prompt: Flat cartoon illustration of a cute coral-red lobster mascot at a podium with a white speech bubble containing ‘International’ text, simplified teal world map glowing in background, dark charcoal background, minimal flag icons scattered at edges, clean sans-serif typography, kawaii mascot style, high contrast, landscape format
Voxtral transcribes at the speed of sound. | Mistral AI https://mistral.ai/news/voxtral-transcribe-2
🎉 Congrats to @Alibaba_Qwen on releasing Qwen3-Coder-Next — and day-0 support is ready in vLLM 0.15.0! An 80B MoE with only 3B active params, matching models 10-20x larger. Built for coding agents and local development. Verified on NVIDIA GPUs. Recipe below 👇”” https://x.com/vllm_project/status/2018742511502856568
🚀 Introducing Qwen3-Coder-Next, an open-weight LM built for coding agents & local development. What’s new: 🤖 Scaling agentic training: 800K verifiable tasks + executable envs 📈 Efficiency-Performance Tradeoff: achieves strong results on SWE-Bench Pro with 80B total params and”” https://x.com/Alibaba_Qwen/status/2018718453570707465
Qwen releases Qwen3-Coder-Next. 💜 The new 80B MoE model excels at agentic coding & local use. With 256K context, it delivers similar performance to models with 10-20× more active parameters. Run on 46GB RAM or less. Guide: https://t.co/kFrY9qi5co GGUF: https://x.com/UnslothAI/status/2018718997584474191
🏆 Agent-Centric Benchmark Results 🟣 SWE-Bench Verified: Qwen3-Coder-Next >70% with the SWE-Agent scaffold 🟣 Efficient but strong: Despite a small active footprint, it matches or exceeds several much larger open-source models on a range of agent benchmarks”” https://x.com/Alibaba_Qwen/status/2018719026558664987
@finbarrtimbers DO NOT use FireworksAI to benchmark Kimi – They have failed to make any of it work right, tool calls aren’t parsed, model is shot up somehow in other ways”” https://x.com/Teknium/status/2018092504613285900
Building a sovereign enterprise https://www.ibm.com/think/insights/ceo-mandate-building-sovereign-enterprise-ai-era?p1=Display&p2=439289718&p3=247627917
🎙️Inside a Chinese AI Lab: How MiniMax Builds Open Models https://www.turingpost.com/p/olive
Congrats to @MistralAI on releasing Voxtral Mini 4B Realtime! 🎉 Day-0 support in vLLM! A 4B streaming ASR model achieving <500ms latency while matching offline model accuracy, supporting 13 languages. vLLM’s new Realtime API `/v1/realtime` provides audio streaming – optimized”” https://x.com/vllm_project/status/2019106596794814894
Heya @FireworksAI_HQ you do not have tool parsing correctly setup for Kimi, please fix or take down the kimi endpoint on openrouter your breaking a lot of my workflows so I had to ban you :)”” https://x.com/Teknium/status/2018155345030627600
joke-generator https://jokegen.sdan.io/blog
Kimi K2 doesn’t listen to system prompt very well. I told it to only generate mermaid charts, not ascii charts, but it keeps generating ascii charts.”” https://x.com/QuixiAI/status/2018213058284229083
Kimi K2.5 | Don’t Worry About the Vase https://thezvi.wordpress.com/2026/02/04/kimi-k2-5/
🚨BREAKING: Kimi K2.5 by @Kimi_Moonshot is now the #1 open model in Code Arena! In Code Arena’s agentic coding evaluations, Kimi K2.5 is now: – #1 open model, surpassing GLM-4.7 – #5 overall, on par with top proprietary models like Gemini-3-Flash – The only open model in the top”” https://x.com/arena/status/2018355347485069800
[2601.21337] Qwen3-ASR Technical Report https://arxiv.org/abs/2601.21337
> – 1T param with 512 experts!!!!! (22B active) Reminder that Very High Sparsity is the new meta, we’re well into the post-V3 regime. Qwen, Xiaomi are already doing 512 experts in 80B models, with tiny intermediate dim I think at this rate V4 will have, like, 2048 (16-24 active).”” https://x.com/teortaxesTex/status/2019245564232364231
Qwen https://qwen.ai/blog?id=qwen3-coder-next
Qwen-Image-Edit-3D-Lighting-Control app, featuring 8× horizontal and 3× elevational positions for precise 3D multi-angle lighting control. It enables studio-level lighting with fast Qwen Image Edit inference, paired with Multi-Angle-Lighting adapter. Try it now on @huggingface.🤗”” https://x.com/prithivMLmods/status/2019084493210992884
Leave a Reply