Aion 1.0 Instruct 2.3B vs Inkling 2 Mini
Compare documented capabilities and access. A dash means a comparable figure is unavailable.
| Specification | ||
|---|---|---|
| Task | On-device language generation | Multimodal reasoning and coding |
| Published capabilities | Windows-oriented instruction model for local text generation, including document and email tasks. | Reuses its middle layers three times per token. Thinking Machines reports 48.2% on DeepSWE v1.1. |
| Model size | 2.3B parameters | 26.4B total / 3.4B active parameters |
| Access | Downloadable checkpoint | Downloadable weights and Tinker inference and customization |
| Operating limits | 32K-token context window | Native mixed-precision checkpoint occupies approximately 20 GB. |
| Sources | Model weights | Model weights · Additional documentation |
| Inputs | Text | Text, Images |
| Weights | Open weights · MIT | Open weights · Apache-2.0 |
| Released | Nov 18, 2026 | Nov 20, 2026 |