TRAE's anniversary is coming up on 1/20! Here's our gift for you: Limited-time Bonus Fast Requests 🎁 1️⃣ Free users get 600 bonus Fast Requests, valid through 02:00 UTC, February 14, 2026. 2️⃣ Pro users get 800 bonus Fast Requests, valid through 02:00 UTC, March 14, 2026. All models can be accessed by applying the bonus during this period. Happy Building with TRAE! Claim here: trae.ai/2026-anniversary-gif…

We just flipped the scaling narrative: Agentic Density > Parameter Count. #MiroThinker 1.5 operationalizes Interactive Scaling—agents that seek evidence, iterate, and revise in real time (with a time-sensitive sandbox to avoid hindsight leakage). Result: 30B hitting frontier-class agentic search, ~$0.07/query (≈20× cheaper vs 1T-class baselines). Fully open source, read more: research.miromind.ai/blog/in… Try: dr.miromind.ai GH:github.com/MiroMindAI/MiroTh… HF: huggingface.co/miromind-ai

Introducing Atoms: the first AI team that builds real businesses. From research to build, launch, and scale, all autonomous. Don’t Vibe Code. Vibe Business. → atoms.dev

This robot assistant from the NVIDIA CES Keynote on Monday is going viral. @NaderLikeLadder explains all the hottest emerging AI trends in one demo: AI applications in 2026 will be multi-model, multi-modal, hybrid cloud/local, use open source models as well as proprietary models, control robots and embedded devices in the physical world, and have voice interfaces. (And the demo had a cute robot *and* a cute dog. Gold.) The demo was built with @pipecat_ai. NVIDIA posted a really nice technical walk-through and complete code. The Reachy Mini robot from @huggingface is open source hardware. (You can order it now, I have one!). You can run the assistant locally on your own hardware, in the cloud, or both.

Miromind.ai is today releasingMiroThinker 1.5. This By matching or exceeding the performance of 1T (trillion) parameter models with a highly optimized 30B architecture, MiroThinker 1.5 validates a new scaling law:Interactive Scaling

Introducing UltraEval-Audio v1.1.0: A Unified Evaluation Framework for Audio Foundation Models with One-Click Reproduction ✅ True One-Click Reproduction: Stop guessing hidden parameters. We provide standardized, "white-box" evaluation pipelines for top models like VoxCPM, MiniCPM-o 2.6, CosyVoice3, GLM-TTS, Kimi-Audio, & Qwen3-Omni, ensuring your local runs align perfectly with official paper results. ✅ Isolated Inference Engine: Model deployment in isolated sandboxes, communicating with the main evaluation process via IPC—enabling safe, parallel inference without dependency conflict . ✅ Expanded Model Coverage: We’ve expanded beyond General Audio foundation models to support specialized domains with tailored benchmarks: - TTS: Focused on task diversity (Seed-TTS-Eval, Long-TTS) including Voice Cloning & long-form synthesis. - ASR: Robust testing across clean, noisy, & multilingual scenarios (LibriSpeech, WenetSpeech). - Codec: A 3-Dimensional system measuring Semantic (WER), Timbre (SIM), & Acoustic (UTMOS) quality. 🔗 GitHub: github.com/OpenBMB/UltraEval… 📄 Paper: github.com/OpenBMB/UltraEval… 📒 Guide: github.com/OpenBMB/UltraEval… #UltraEval #AI #opensourceai #MiniCPM #TTS #ASR #OpenBMB