TypeLLM Forces LLMs to Return Valid JSON Every Time, 5.8x Faster
TypeLLM constrains open-source LLM decoding to your JSON Schema, delivering guaranteed typed outputs with one-token choices and prefix-cached batching.
20 stories
TypeLLM constrains open-source LLM decoding to your JSON Schema, delivering guaranteed typed outputs with one-token choices and prefix-cached batching.
A 1.5B parameter Qwen2-based anime chatbot lands in imatrix GGUF format, running under 1GB on plain consumer laptops.
Anthropic's Claude autonomously pushed a frontier theoretical physics calculation from eight to nine loops on Claude Science, costing only a few thousand dollars.
Runway's hosted MCP server plugs Gen-4.5, Seedance 2.5, Veo 3.1, Kling and other models directly into Claude, ChatGPT, Cursor and Replit with no API key.
A 9B Qwen3.5 fine-tune strips refusal behavior and adds agentic terminal skills, targeting local red-team workflows with GGUF quants that fit on a laptop.
Exa Agent Ultra orchestrates swarms of subagents to run exhaustive web research, topping WANDR, WideSearch and DeepSearchQA at lower cost than frontier rivals.
Replit is folding Atta's purpose-built charting engine into its agent, letting anyone generate interactive business charts inside chat without writing SQL.
Zed's new agent harness Delta ties or beats Codex and Claude Code on Terminal-Bench accuracy at 0.80 to 1.12 times the cost per pass.
Rome is an open-source agentic OS where humans and agents build reusable apps, workflows, and capabilities that compound over time.
PR Lens turns pull requests into animated architecture and data-flow diagrams that show the shape of a change before you read a single line of code.
Runway added a one-click tool that decomposes any image into transparent, editable layers using ByteDance's Seedream 5.0 Layers model, no manual masking required.
Moondream shrank NVIDIA's Parakeet speech model from 1.2GB to 178MB using ternary weights, hitting 113x realtime on CPU while matching accuracy.
Together AI released a 4B decision classifier fine-tuned from Qwen3.5, trained for just $17, with a full recipe to build your own.
Fastino's 340M encoder classifier scores typed decisions in one forward pass, beats a 4B baseline on 17 tasks, and runs on CPU.
A cross-platform rewrite of Wifite ships its own userland USB drivers, sidestepping kernel headaches on both Linux and Windows
Sarvam's new speech-to-text model hits state-of-the-art on all 22 Indian languages plus seven English benchmarks, with five output modes in one model.
A distilled predictor skips 90% of attention tiles in MiniMax H3, delivering up to 2.9x faster text-to-audio-video generation in ComfyUI without touching model weights.
Ornith-1.5-9B lands in NVFP4 format, delivering a Blackwell-native 4-bit reasoning model that punches near 35B-class agents at a fraction of the memory.
A new smoothing trick lets differentiable simulators keep stiff, realistic contact physics while still producing gradients stable enough to train humanoid policies that transfer zero-shot to a real Unitree G1.
Naver's sparse retrieval model hits 878K downloads as teams rediscover SPLADE for hybrid search pipelines that need lexical precision plus neural understanding.