Thinking Machines has released Inkling 2 Mini, a looped transformer model designed to run on consumer laptops and desktop AI computers. It excels in tasks requiring reasoning and is optimized for speed and efficiency.
Key facts
| Fact | Detail | The source says |
|---|---|---|
| Total Parameters | 26.4B | “Params (B) activated / total / 5.4/26.4” |
| Memory Footprint | ~20GB | “The total memory footprint of Inkling 2 Mini is ~20GB” |
| DeepSWE v1.1 Score | 48.2% | “Inkling 2 Mini leads every open-weights model with less than 100B total parameters on benchmarks such as DeepSWE v1.1 (48.2%)” |
What happened
Thinking Machines has released Inkling 2 Mini, a looped transformer model designed to run on consumer laptops and desktop AI computers. It excels in tasks requiring reasoning and is optimized for speed and efficiency. Inkling 2 Mini is the first model with a looped architecture available to researchers, allowing for cleaner comparisons between different looping depths. The model is small enough to run on local devices, with a total memory footprint of ~20GB. It leads open-weights models of similar size on benchmarks such as DeepSWE v1.1 (48.2%) and Vals Index v2.2 (46.5%). Inkling 2 Mini is available for inference and customization on Tinker and supports various platforms for local inference.
What to weigh
- Inkling 2 Mini is optimized for local hardware and runs efficiently on NVIDIA’s DGX Spark desktop AI computer.
- The model is designed for tasks requiring reasoning and is the best among open-weights models of similar size on most tasks.
Source: thinkingmachines.ai

