Cohere Labs' North Small Translate Beats DeepL and Google Translate Across 50 Languages
Cohere's new 218B MoE translation model tops WMT26 against DeepL, Google Translate, and open alternatives across 50+ languages under a non-commercial license.
// Intelligence
Open Source
A focused feed of models, agents, research and open-source releases for people building with AI.
Updated
Tencent's AuK Replaces 16 Speech Tools With One Open-Source Model
Tencent Hunyuan opens the code and weights for a 1.5B unified speech model that handles TTS, editing, denoising, and separation from plain instructions.
Unitree's UnifoLM-WLA-1.0 Runs 64 Robot Tasks From One Set of Weights
Unitree open-sourced a 6B-parameter humanoid foundation model that runs 64 tasks with one network across grippers, dexterous hands, and full-body motion.
CozyClay Lets You Direct AI Video Shots in 3D Without Writing Prompts
CozyClay turns your browser into a previs studio where you block shots, pose characters, and hand the same camera moves to an AI video model.
YuE2-3B Beats Suno Generating Editable Full Songs on a Single GPU
An open 3B music model from M-A-P matches Suno v5 on quality, adds editable ABC scores, and runs on a 24GB GPU.
Open-Source Wake Unifies Claude Code and Gemini CLI Histories on macOS
Wake is a native macOS app that unifies coding-agent sessions from Claude Code, Codex, Cursor and nine more into one searchable window.
Cohere's Open-Source Megakernel Beats vLLM by 1.58x on H100
Cohere open-sourced a serving engine that runs the entire LLM decode step as one persistent CUDA kernel, hitting 1.58x vLLM throughput on H100.
HarnessRouter Unifies Codex, Claude Code, and Hermes Behind One Open API
An open-source Docker container runs Codex, Claude Code, and Hermes through one unified API, keeping your keys and data on your own machine.
vLLM's Hybrid HiSparse Triples Concurrent Requests on Million-Token Contexts
vLLM's new Hybrid HiSparse keeps long-context requests decoding when the KV cache overflows HBM, tripling concurrency on GLM 5.3 at 1M context.
cmpunlocker Turns NVIDIA's Dead Mining GPU Into a 64GB Powerhouse
An open-source patch restores 64GB of HBM2e memory and full compute on NVIDIA's crippled CMP 170HX mining card, transforming a $200 crypto relic into a viable AI accelerator.
Adaption Labs' AutoScientist Ships One-Click Hugging Face Exports for Fine-Tuned Models
Adaption Labs plugs its automated fine-tuning system into Hugging Face, letting anyone push a co-optimized model to the Hub with a single click.
Nex-AGI Drops Nex-N2.5, a 397B Open Agent That Controls Your Desktop
Nex-AGI releases an open-source agentic model family with vision-driven computer use, three sizes, and benchmark scores rivaling Claude Opus 5.
AI-FanGe's Microduck Lets You 3D Print a Walking Robot for Under $400
A hobbyist published a full DIY build kit for a Microduck-style RL biped robot, complete with a prebuilt SD image and MuJoCo training environment.
RadixArk's Miles v0.1 Cuts Trillion-Parameter RL Sync Time by 86%
RadixArk's open-source Miles framework runs asynchronous RL on a 744B-parameter model across 64 GB300 GPUs with a 263-second median step.
Nex-AGI Releases Nex-N2.5 Open Agents That Browse, Click, and Code
Nex-AGI open-sourced a three-tier agentic model family under Apache 2.0, topped by a 1.6T-parameter MoE built to run browsers, desktops, and code loops end to end.
Red Hat's Ripwire Cuts Coding Agent Token Use by 96% Without Embeddings
Red Hat's ripwire is a zero-dependency C++23 CLI that hands coding agents a ranked, deterministic map of any repo without embeddings or vector DBs.
AIPOCH's Open Science Gives Researchers a Local AI Agent With 4,000 Stars
AIPOCH released Open Science, an Apache-2.0 desktop workbench that runs scientific agents locally with Python notebooks, life-science connectors, and full provenance.
UC Berkeley's VDN-H3 Generates 14-Second Videos Faster Than You Can Watch
VDN-H3 replaces most of MiniMax H3's quadratic attention with a linear branch, rendering a 14.4-second 768p clip in 11.23 seconds on 8 B200s.
Community LoRA Sharpens MiniMax H3 Videos Without Hosted 2K API
A community LoRA for MiniMax H3 sharpens video output through aligned guide latents, offering a new pass on ComfyUI's ref2va pipeline.
GLM-5.3 Gets Its Safety Stripped, Hits 89% Compliance on Cyber Attacks
A 753B parameter GLM-5.3 variant has had its refusal circuits surgically removed for offensive security work, keeping MMLU intact while complying with 89% of cyber-attack prompts.
Zhijian Skills Bundles 16 Open-Source Agent Workflows Into One Catalog
A Chinese developer's open-source Agent Skills portfolio bundles sixteen focused tools, including a layered Web Clipper that chains five extractors to reliably scrape anything.
Experiential Gives AI Agents One Open-Source Gateway to Rule All Models
Experiential is an open source, OpenAI-compatible gateway that routes across hosted, BYOK, and local models with zero markup and spend controls.
Alibaba's zvec-grep Gives Coding Agents One Search Layer to Replace Endless Grep Loops
Alibaba's Zvec team open-sourced zg, a local-first CLI that fuses ripgrep, BM25, and vector search into one interface for humans and coding agents.
Perplexity's Lily Beats MLX-LM by 1.35x Running Qwen3.6 on Apple Silicon
Perplexity open-sources Lily, a Metal-based inference engine tuned for Qwen3.6-35B-A3B that beats MLX-LM by 1.23x prefill and 1.35x decode on M5 Max.
No stories match these filters.