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.
Thinking Machines
Frontier AI research and model company focused on novel architectures and inference systems, developer of the Inkling model fami
Curated Intelligence
Thinking Machines has released Inkling 2 Mini, a 26B-parameter looped transformer small enough to run on a laptop yet capable of agentic work, scoring 20.1 on the AA Index. The model is now live across Arena.ai's Agent Arena, Code Arena and Text Arena, with leaderboard results pending.
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.
Thinking Machines released Inkling 2 Mini, a multimodal MoE with 26B total and 3.5B active parameters, up to 1M token context, and a native DSpark speculator. Its key innovation is a looped architecture where the middle 12 layers run three passes with shared weights, each keeping its own KV cache and reading gated messages from the prior pass. vLLM shipped day-0 support on NVIDIA Blackwell, Hopper, and DGX Spark.
Thinking Machines has released the weights of Inkling 2 Small on HuggingFace under Apache 2.0. The model scores 1647 on Arena.ai's Code Arena WebDev leaderboard, ranking first among US open models, fourth among open models globally and 23rd overall, a 239-point gain over its predecessor Inkling-Small at 1408.
vLLM shipped day-0 support for Thinking Machines' Inkling 2 Small, a multimodal MoE model with 594B total and 17B active parameters and up to 1M tokens of context, bundled with a native DSpark speculator. Through kernel fusion, PDL, a tuned Gluon attention kernel and sequence parallelism, vLLM reaches up to 630 TPS per user at TP4 on GB200 with acceptance length 3.5, while also supporting LoRA and RL workloads.