OpenRouter: from Seed to Stripe — with OpenRouter’s Alex Atallah & AMP’s Anjney Midha
15 min
Liquid AI Releases LFM2.5-VL-3B-DSpark: Speculative Decoding for Vision-Language Models With Up to 3.13x Faster Decoding
5 min
Best LLM for every budget, updated daily
2 min
Mercury 2.5 LLM hits 770 tokens per second
6 min
Memory Attention
Tokens Too Cheap to Meter
14 min
Better prompt caching for GPT-6
3 min
FLEET: From Logits Entropy to Enhanced Trajectories in Text Generation
Flash-dLLM: IO-Aware KV Caching and Parallel Decoding for Fast, Memory-Efficient Diffusion LLMs
Transformers now runs llama.cpp quants
M5 Ultra Mac Studio Review: The Dream Mac for Local AI Agents
14 min
Cache-to-Cache: Direct Semantic Communication Between LLMs (2025)
1
Watermarkable Multi-Draft Speculative Sampling via Poisson Processes
Alibaba releases Qwen 3.8 Omni Flash
GLM Built Its Own Inference Infrastructure
Former OpenAI researcher builds an AI model that judges options instead of writing text
3 min1
NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 Debut
6 min1
harshatheg/Qwen-2.5-1B-RLCD — new model trending #30 on Hugging Face
9 min2
The Other Half of the Memory Wall: Serving 35B MoEs from SSD with Trained Routing Prediction
Jev: New frontier model 40-400x cheaper and 20-200x faster
10 min
JustFit: 200K-Token LLM Serving on a 24 GiB Laptop with Just-in-Time State Management
FlashVector: Agent for Hierarchical Model Serving Stack Optimization
1
Show HN: Sunk Cost – How long until a local LLM rig pays for itself?
Show HN: Nari Qwen3-TTS and Qwen3-ASR – High accuracy, low latency and cost
3 min1