By olud.ai editorial · built from our own tracking data, published daily
Docling v2.127.0 shipped with three new features: region support in S3Coordinates, an MHTML input backend, and RTF input support via LibreOffice. Open Interpreter rust-v0.0.43 arrived as a maintenance release based on Codex rust-v0.154.0, adding easy choices for Gemini 3.8 Flash and 3.7 Flash, Z.AI ZCode and GLM support.
The ecosystem gained 57,684 GitHub stars across tracked projects today, with 78 new projects entering the index. This week 2,121 new projects and 100 new Hugging Face spaces were added. The top gainer today is "archify" with +3,433 stars.
Releases of the day
Docling v2.127.0 expands document input options with MHTML and RTF support, making it easier to convert web archive and rich text files. The addition of region to S3Coordinates improves storage location handling. Open Interpreter rust-v0.0.43 updates the code interpreter with support for newer Gemini models, Z.AI's ZCode, and GLM, giving users more model choices for interactive coding tasks.
DeepSeek V4 Flash 0423 input price fell 29% to $0.05 per 1M tokens. Qwen3 30B A3B Instruct 2507 dropped 44% to $0.05 per 1M tokens. GLM 5.3 from Z.AI decreased 22% to $1.09 per 1M tokens. On the other side, Gemma 4 26B A4B input price rose 125% to $0.09 per 1M tokens, and Muse Glimmer 30B increased 17% to $0.35 per 1M tokens. These moves affect users of these models on their respective platforms.
Among the 78 new projects tracked today, "archify" leads with over 3,400 GitHub stars. The project's rapid growth suggests strong early interest, though details on its functionality are not provided in today's data.
Research pick
A new paper titled "Benchmark Radar: A Living Database and Search Engine for AI Benchmarks and Evaluation" proposes a tool for researchers and developers to find relevant evaluations, locate benchmark datasets and code, and understand the settings behind reported scores. The work aims to simplify navigating the growing landscape of AI benchmarks.
Today's facts show a market where model prices are diverging sharply, with several providers cutting costs while others raise them. New document format support in Docling and expanded model choices in Open Interpreter point to continued tooling improvements, while the steady influx of new projects and research tools like Benchmark Radar reflect a community focused on making AI evaluation more accessible.
Source: olud.ai tracking of 10,000+ open-source AI projects, 300+ models and live provider pricing. Figures are measured, not estimated. All releases · Live pricing · Latest in AI
