Vladislav Shter, solo-founder of the Sovereign Ecosystem, has developed Egregor, a desktop application that simultaneously runs multiple AI models to audit smart contracts and code. In testing, Egrego…
Developer Youssef Mejdi released Grinta v1.0.0-rc1, a local-first, provider-agnostic coding agent for long autonomous runs that integrates with LSP and DAP. The tool runs entirely on the user's machin…
A developer built a $0 local AI automation stack combining agentic coding and market data analysis using Aider, Ollama, OpenBB, and MCP servers. The stack runs entirely on free and local infrastructur…
TradingSpy, an open-source, local-first AI trading assistant, has been released. The tool provides market heatmaps, news catalysts, strategy generation, and backtesting in a single Docker application,…
A developer open-sourced TradingSpy, a local-first AI trading research workstation that runs entirely in Docker. The platform uses AI agents to autonomously generate and backtest trading strategies, s…
Benchmark partner Peter Fenton predicts open-weight AI will dominate token usage within two years, driven by developer-led adoption and cost pressure. The forecast suggests premium pricing for proprie…
An educational lab demonstrating various AI agent architectures built on LangChain and a local Ollama server has been released. The project includes multiple runnable CLI variants for studying chat wi…
OpenRouter and Andreessen Horowitz analyzed over 100 trillion tokens of real-world LLM usage, finding that open-weight AI models now account for roughly one-third of all token volume. Chinese open-wei…
Inferock released inferock-bench, an open-source local proxy that generates per-call billing receipts for OpenAI, Anthropic, and other LLM providers, enabling users to audit token usage, costs, and fa…
OpenAI's GPT-5.6 Sol, released three days ago, dominates the PR security review benchmark with 100% recall, 83.3% precision, an F1 of 0.91, and a cost of $0.70 per pull request, according to Dam Secur…
DeepSeek releases V4 Pro, a 1.6-trillion-parameter Mixture-of-Experts model with 49 billion active parameters per token, achieving a top open-weight score of 80.6% on SWE-bench Verified. The model req…
Chinese AI models from Zhipu AI and DeepSeek are gaining developer share in the United States due to lower costs, with OpenRouter data showing Chinese model token consumption above 20% weekly share si…
Zhipu AI's GLM-5.2 open-weight model outperformed DeepSeek V4 Pro and Kimi K-2.6 in enterprise agentic development trials, but its limited hosting options—only available through small US, Chinese, and…
Europe's AI strategy relies on open-weight models from China and the US, but access to these models is being restricted by Beijing, Washington, and private labs, leaving the continent without a viable…
Software developer Tarun Gupta created Autopilot-Jobhunt, an AI tool that scans the web for job openings matching a user's resume, scores them, and sends alerts via Telegram. The tool can also generat…
Tencent released Hy3, a 295-billion-parameter open-weights AI model, at roughly six cents per million input tokens, outperforming GLM-5.1 in blind expert reviews. The model features 21 billion active …
DeepSeek released V4 Flash, a 284B-parameter MoE model with 13B active parameters per token, featuring FP4+FP8 hybrid attention and 1M native context. The open-weight model, available under MIT licens…
GLM 5.2, a 744B-parameter Mixture-of-Experts model with ~40B active parameters per token, was released in June 2026 under an MIT license. It uses MLA and DeepSeek Sparse Attention for a 1M context win…
NVIDIA released Nemotron 3 Ultra, a 550-billion-parameter open-weight MoE model with hybrid Mamba-2 and Transformer architecture, under the Linux Foundation's OpenMDW-1.1 license. The model requires s…
Databricks' internal benchmark of AI agents on its multi-million-line codebase reveals that token costs are misleading, agent harness design significantly impacts performance, and open-source models l…