Trending

LLMs Encode How Difficult Problems Are

Show HN: qqqa – a fast, stateless LLM-powered assistant for your shell

Brain-IT: Image Reconstruction from fMRI via Brain-Interaction Transformer

Launch HN: Plexe (YC X25) – Build production-grade ML models from prompts

Tongyi DeepResearch – open-source 30B MoE Model that rivals OpenAI DeepResearch

Context engineering

The Smol Training Playbook: The Secrets to Building World-Class LLMs

Why do AI models use so many em-dashes?

Llamafile Returns

OpenAI's New GPT-5.1 Models are Faster and More Conversational

OpenAI recently released upgrades to their GPT-5 model. GPT‑5.1 Instant, the default chat model, has improvements to instruction following. GPT‑5.1 Thinking, the reasoning model, is faster and gives more understandable responses. GPT‑5.1-Codex-Max, the coding model, is trained to use compaction to perform long-running tasks. By Anthony Alford

Trinity Mini: a 26B OpenWeight MoE model with a 3B active and strong reasoning scores

LLMs can understand Base64 encoded instructions

Z.ai release GLM-ASR-Nano: an open-source ASR model with 1.5B parameters

Introducing: Devstral 2 and Mistral Vibe CLI. | Mistral AI

bartowski/mistralai_Devstral-Small-2-24B-Instruct-2512-GGUF

DeepSeek-V3.2-REAP: 508B and 345B checkpoints

Devstral-Small-2-24B-Instruct-2512 on Hugging Face

Which small model is best for fine-tuning? We tested 12 of them by spending $10K - here's what we found

zai-org/GLM-4.6V-Flash (9B) is here

Microsoft

Save item

VibeVoice-Realtime-0.5B

Task: text-to-speech

Z.ai

Save item

GLM-4.6V-Flash

Task: image-text-to-text

Mistral AI_

Save item

Devstral-Small-2-24B-Instruct-2512

NVIDIA

Save item

Nemotron-Orchestrator-8B

Task: text-generation

ServiceNow-AI

Save item

Apriel-1.6-15b-Thinker

Task: image-text-to-text

Apple

Save item

CLaRa-7B-Instruct

Mistral AI_

Save item

Ministral-3-14B-Instruct-2512

NousResearch

Save item

Hermes-4.3-36B

Task: text-generation

Microsoft

Save item

VibeVoice-1.5B

Task: text-to-speech

ggml-org

Save item

whisper.cpp

Port of OpenAI's Whisper model in C/C++

openai

Save item

whisper

Robust Speech Recognition via Large-Scale Weak Supervision

huggingface

Save item

transformers

🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

NVIDIA

Save item

Megatron-LM

Ongoing research training transformer models at scale

f

Save item

awesome-chatgpt-prompts

This repo includes ChatGPT prompt curation to use ChatGPT and other LLM tools better.

huggingface

Save item

pytorch-image-models

The largest collection of PyTorch image encoders / backbones. Including train, eval, inference, export scripts, and pretrained weights -- ResNet, ResNeXT, EfficientNet, NFNet, Vision Transformer (ViT), MobileNetV4, MobileNet-V3 & V2, RegNet, DPN, CSPNet, Swin Transformer, MaxViT, CoAtNet, ConvNeXt, and more

Dao-AILab

Save item

flash-attention

Fast and memory-efficient exact attention

run-llama

Save item

llama_index

LlamaIndex is the leading framework for building LLM-powered agents over your data.

deepspeedai

Save item

DeepSpeed

DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.