# MOE: what I've logged

> MOE comes up in 2 posts of my AI log, from October 2026 to November 2026, and each item below links to the primary source. The sources come from Thinking Machines and inclusionAI, filed under open weights, frontier and science.

## Items

1. **Welcome Inkling 2 Mini** (Thinking Machines, 20 Nov 2026): A multimodal MoE with about 26B total and 3B active parameters, a 1M token context and a looped middle of the decoder, with day-zero support in Transformers, llama.cpp, MLX and SGLang. [Primary source](https://huggingface.co/blog/inkling2mini) · [Post](https://heyaliasif.com/ai/2026-11-20-inkling-2-mini-openai-distillation-medchembench-self-rewriting-weights#item-1)
2. **Ling-3.1-flash** (inclusionAI, 14 Oct 2026): A 560B hybrid-linear MoE with 25B active parameters and a 1M-token window, MIT licensed, with day-0 SGLang support and multi-token-prediction decoding. [Primary source](https://huggingface.co/inclusionAI/Ling-3.1-flash) · [Post](https://heyaliasif.com/ai/2026-10-15-jais-3-step-5-ling-drama-3#item-3)

## Posts

- 2026-11-20 [Inkling 2 Mini, OpenAI on distillation, MedChemBench](https://heyaliasif.com/ai/2026-11-20-inkling-2-mini-openai-distillation-medchembench-self-rewriting-weights)
- 2026-10-15 [Jais 3 goes open, Step 5 and Ling 3.1 land, Drama 3 speaks](https://heyaliasif.com/ai/2026-10-15-jais-3-step-5-ling-drama-3)

Canonical: https://heyaliasif.com/ai/topics/moe