Termsslug: looped-transformer

Looped Transformer

循环 Transformer

Architecture that reuses the same layer weights across multiple passes to add effective depth without more parameters.

Curated Intelligence

12 items · Page 1/2
Marktechpost (@Marktechpost)T2·Models
78 pts
Thinking Machines Open-Sources 26.4B Looped Transformer Inkling 2 Mini

Thinking Machines released Inkling 2 Mini, a looped transformer that stores 24 layers but runs the middle 12 three times with shared weights, yielding 48 layer passes per token from a ~20GB checkpoint that fits on a 32GB Mac. The 26.4B-total, 3.4B-active model supports 1M context under Apache 2.0, scoring 79.2% on SWE-bench Verified and 48.2% on DeepSWE 1.1 versus 42.2% for Qwen3.8-27B, while knowledge recall lags notably.

The DecoderT2·Models
78 pts
Thinking Machines Open-Sources 26.4B Looped Transformer Inkling 2 Mini

Thinking Machines released Inkling 2 Mini, an open looped transformer that runs 12 of its 24 layers three times per token, extending internal reasoning without emitting more text. The 26.4B-parameter mixture-of-experts model activates 5.4B parameters per token, handles text and images, offers a 1M-token context window, ships under Apache 2.0, and runs on a laptop. It leads open models under 100B parameters on agentic coding benchmarks, scoring 48.2% on DeepSWE v1.1 and 46.5% on Vals Index v2.2.

机器之心 (微信公众号)T2·Models
78 pts
Thinking Machines Lab Open-Sources 26B Looped Transformer That Runs on Laptops

Thinking Machines Lab has open-sourced Inkling 2 Mini under Apache 2.0, a 26B-parameter multimodal looped transformer supporting text and image inputs with a 1M-token context window, plus inference and fine-tuning via Tinker. In an offline demo on a 48GB M5 Pro MacBook Pro, it generated a playable flight simulator in minutes, scoring 20.1 on the AA Index and targeting agentic workloads on personal hardware.

Artificial AnalysisT1·Models
78 pts
Thinking Machines Releases Inkling 2 Mini Open Weights Model

Thinking Machines released Inkling 2 Mini, a 26B total / 5.4B active parameter mixture-of-experts model built on a looped transformer architecture, open-sourced under Apache 2.0 with text and image input and a 1M token context window. It scores 20 on the Artificial Analysis Intelligence Index, matching Qwen3.8 27B and Nanbeige4.3-33B with roughly 5x and 6x fewer active parameters respectively, sits on the Pareto frontier for intelligence versus size, and is small enough to run on a MacBook or DGX Spark.

Artificial Analysis (@ArtificialAnlys)T1·Models
78 pts
Thinking Machines Releases Inkling 2 Mini Open-Weights Model

Thinking Machines has released Inkling 2 Mini, a 26B-total, 5.4B-active-parameter looped-transformer mixture-of-experts model scoring 20 on the Artificial Analysis Intelligence Index, placing it on the Pareto frontier for intelligence versus active parameters. Released under Apache 2.0 with text and image input and a 1M-token context window, it is small enough to run locally on a MacBook or DGX Spark.

Hacker News AIT2·Models
74 pts
Inkling Releases 26.4B Looped Transformer Model That Runs on Laptops

Inkling released Inkling 2 Mini, an open-weights model built on a looped transformer that runs its middle layers three times per token to reason with fewer tokens. The MoE model has 26.4B total and 3.4B active parameters, scoring 48.2% on DeepSWE v1.1 and 46.5% on Vals Index v2.2, which it claims leads all open-weights models under 100B while running locally on a laptop.

Hugging Face BlogT1·Models
74 pts
Thinking Machines Ships Inkling 2 Mini Looped Transformer for Local Setups

Thinking Machines Lab and Hugging Face released Inkling 2 Mini, a multimodal Mixture-of-Experts model for local deployment with roughly 26B physical and 3B active parameters per token. Its key trick is looping certain layers three times to boost effective capacity, alongside a 1M-token context, image and text input, and mixed-precision NVFP4/MXFP8 expert weights with a DSpark drafter for extra speed.

Hacker News AIT2·Models
68 pts
Tiny World Model and RL Agent Run on Raspberry Pi Pico 2 W

A developer deployed a tiny world model based on looped transformers on a Raspberry Pi Pico 2 W microcontroller, enabling real-time game frame prediction and interaction with an RL agent. The approach significantly reduces memory usage through weight reuse, demonstrating the feasibility of running generative AI models on edge devices.