{"slug": "show-hn-open-source-daidocs-now-beats-graphify-by-218x-low-tokens-83-on-lme", "title": "Show HN: Open-Source DaiDocs now beats Graphify by 218x low tokens, 83% on LME", "summary": "The open-source DaiDocs project released a new .dai (Document AI) and .cai (Code AI) file standard that it says cuts token usage by more than 10x versus raw files on LongMemEval-S and by 218x versus Graphify on coding-agent tasks, with reproducible numbers posted at github.com/kerneta/daidocs. DaiDocs reports its .dai format scored 83% with GPT-4 and 92% with Fable 5 on long-conversation memory and can fit more than 1 million tokens of chat into a model with a 1 million-token context window. The project urges community contributions, arguing no vendor such as Anthropic or OpenAI would develop the standard itself.", "body_md": "We have been working on making an Open-Source standard documents for AI LLM's and Agents. Just like an excel file, .dai (Document AI) is a new standard to keep memory local and use with any AI agents and LLM's. This new standard comes with lots of features such as more than 10x tokens savings (tested on LongMemEval-S) when comapred the .dai to raw files and scored 83% with gpt-4-0 and 92% with Fable 5! (This shows how well it remembers a very long conversation), as well as an increased context window (fitting more than 1m tokens chat in a model which has a max of 1m context window) and much more. Then we moved to coding agents and make the .cai (Code AI) file standard and when compared to Grpahify, these savings were 218x and compared to a raw file it were 2976x! That's a LOT! We have all the codes and reproduceable numbers on our github.com/kerneta/daidocs. We urge the community to try it and contribute as much as they can, as no vendor (like Anthropic or OpenAI) would ever develop this and this is all community driven!\n\nComments URL: [https://news.ycombinator.com/item?id=50004719](https://news.ycombinator.com/item?id=50004719)\n\nPoints: 2\n\n# Comments: 0", "url": "https://wpnews.pro/news/show-hn-open-source-daidocs-now-beats-graphify-by-218x-low-tokens-83-on-lme", "canonical_source": "https://github.com/Kerneta/daidocs", "published_at": "2026-10-08 11:50:56+00:00", "updated_at": "2026-10-08 12:18:19.925676+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "ai-tools", "developer-tools"], "entities": ["DaiDocs", "Graphify", "LongMemEval-S", "GPT-4", "Fable 5", "Anthropic", "OpenAI", "github.com/kerneta/daidocs"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/show-hn-open-source-daidocs-now-beats-graphify-by-218x-low-tokens-83-on-lme", "markdown": "https://wpnews.pro/news/show-hn-open-source-daidocs-now-beats-graphify-by-218x-low-tokens-83-on-lme.md", "text": "https://wpnews.pro/news/show-hn-open-source-daidocs-now-beats-graphify-by-218x-low-tokens-83-on-lme.txt", "jsonld": "https://wpnews.pro/news/show-hn-open-source-daidocs-now-beats-graphify-by-218x-low-tokens-83-on-lme.jsonld"}}