Intel / article

Latest Open Artifacts (#23): Laguna S2.1, Inkling, & Kimi K3 Show the Utility of Open Models on the Pareto Frontier

Source: interconnects.ai
Author: Florian Brand
Date: Published Aug 2, 2026

Context

Open-model releases can differ in how much evidence they expose about their behavior. This roundup highlights Poolside for publishing evaluation runs alongside Laguna S 2.1, giving developers something more informative than a score alone.

The Laguna S 2.1 release provides a concrete example. Poolside says it publishes complete run histories for every trial in its final benchmark evaluation set. Developers can inspect the actions behind a result, while the release's stated limitations identify problems such as mishandling slightly different tool interfaces. That makes the release useful for assessing how a model works inside an , beyond comparing headline scores.

Key takeaways · AI-distilled

Thinking Machines' Inkling is a 975B-A41B model taking text, images and audio in and text out. It isn't the strongest in its size class; it's built to be fine-tuned, feeding the company's paid tuning service Tinker.
Poolside's Laguna-S-2.1 is a 118B-A8B MoE sized to fit on a single DGX Spark desktop box, and that fit is most of why it got . Poolside also published the full evaluation trajectories, which open releases almost never do.
OpenMDW is an Apache-2.0-style free license written specifically for model weights. MIT and Apache are software licenses, so their legal footing over weights and data is shaky; OpenMDW closes that gap. NVIDIA and Poolside both adopted it.
Tencent's Hy3 (295B-A21B MoE) dropped the restrictive custom license its predecessor shipped under and went Apache 2.0, converting a look-but-don't-touch release into one you can build a commercial product on.
Meituan's LongCat-2.0, a 1.6T-parameter MoE, was trained entirely on Huawei Ascend 910s. Chinese accelerators are normally used only for , so this is the first serious model trained end to end without NVIDIA silicon.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
mixture-of-experts — A model built from many specialist sub-networks where only a few activate per token, giving big-model capability at small-model running cost.
attention — The mechanism that lets a model weigh which earlier words matter for the word it's currently processing — the core operation of a transformer.
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.

Key quotes

“But if a US company needs a contract with Moonshot to provide the inference tokens that Kimi K3 generates, the picture looks different.”
Kevin Xu and Graham Webster

“Some of the policy tools US officials and others have debated as potential levers to restrict Chinese open model use would more clearly apply.”
Kevin Xu and Graham Webster

“This is a lot of transparency for an open model release!”

Read the source www.interconnects.ai