Open Source Factory

Mac Studio 512GB: Can It Run a 1T-Parameter AI Model? Open Source Factory · October 29, 2026 · Source-based capacity analysis Watch the English motion graphic An editorial animation of checkpoint capacity, not a Mac benchmark recording. Motion, newly composed original music and sound design: Open Source Factory. A trillion-parameter coding model on one Mac was the first thing I wanted to check. Kimi K2.7 Code makes that a plausible project: one published Q3 package, including its vision projector, is about 419.8GB . The catch is in the word “package.” That number leaves out the operating system, working buffers and conversation cache. The 512GB Mac Studio opens a serious door for giant local AI. I narrowed this comparison to five models with roughly 480 billion to one trillion parameters, rather than filling it with smaller models that miss the point. I checked downloadable weights, exact file sizes and documented runtime routes. This is a capacity and compatibility analysis; I have...

MODELS & COSTS Muse Spark Open: The Memory Behind a Two-Mac Setup Open Source Factory · Oct 18, 2026 · 6 min read Watch the English motion graphic A 48-second English editorial animation. Figures are sourced or calculated, not a recorded inference test. Motion graphics and newly composed original music, “Shared Memory”: Open Source Factory. Two Mac Studios sound like a compact home for an 816B model. The configuration matters: Meta's documented recipe uses specific memory capacities, a mixed-precision checkpoint and MLX. A count of two computers is not enough to price or reproduce that setup. In this article What did Meta actually release? Why does a 42B-active model need so much memory? Can you buy the two-Mac setup today? What if you do not own a terabyte of memory? Does “open” mean you can use it without conditions? Verify the complete Mac configuration Meta documents two M5 Ultra Mac Studios with 512GB each , or four with 256GB each , using its mixed-precision NVFP4 check...

MODELS & COSTS Reflection Beam: Store the Whole Model, Not 23B Open Source Factory · Oct 05, 2026 · 6 min read Beam's active parameter count is 23B, while its total is 501B. Routing can reduce the work performed for a token without removing the other experts from storage. That is why the memory calculation starts with 501B, before discussing a computer to host it. In this article Release status at the analysis date Active computation versus stored weights Raw payload at different precisions Inference claims and the coding bill On October 5, Reflection introduced Beam , a sparse Mixture-of-Experts model aimed at coding and reasoning. As of October 6, its public weights are still due later this month. My storage calculation is already sobering: 501B weights at a hypothetical uniform 4 bits each take 250.5GB before any runtime overhead. Image: Reflection’s Beam announcement . Official artwork; the circles do not specify the model’s routing architecture. Release status at the anal...

MODELS & COSTS MiMo V2.6 Pro: The Task Cost and the Training Record Open Source Factory · Sep 21, 2026 · 12 min read Xiaomi MiMo — the V2.6 series went up on Hugging Face under the MIT license. (image: Xiaomi) MiMo's low measured evaluation cost makes a trial affordable. Xiaomi's published training records add something different: a way to inspect how the release was developed, including reported failures. Neither a cheap test nor a detailed log establishes that it can replace Opus on your own work. In this article The Pro and Flash releases The independent score of 46 Rows outside the launch headline The published six-day training record Results attributed to the training run Examples outside coding Prices and measured task costs Testing an Opus workload on MiMo Here is the part that made me stop scrolling. MiMo-V2.6-Pro scores 46 on the Artificial Analysis Intelligence Index, which makes it the highest-ranked open-weights model on the board — one point ahead of GLM...

MODELS & COSTS Qwen-Image-2.1: A Cheap GPU Estimate Has License Conditions Open Source Factory · Sep 20, 2026 · 10 min read Rented GPU time can make a 10,000-image estimate look inexpensive. That does not buy commercial permission, matching image quality or a finished workflow. Qwen-Image-2.1's research license and the hardware assumptions belong beside the calculation from the beginning. In this article Release capabilities and integrations The components behind the 7B label Memory and dtype requirements A rental-cost estimate for 10,000 images The research-license restriction Quality comparisons and reported failures The release uses the Qwen Research License Agreement, limited to non-commercial research and evaluation in the cited terms. Earlier open Qwen image releases used Apache 2.0. Downloadable weights and familiar tool support do not remove that licensing difference. The model card the day after launch: 888 likes, and the licence tag reads "other" rathe...