CCORE Web Coder 4B vs Inkling 2 Mini
Compare documented capabilities and access. A dash means a comparable figure is unavailable.
| Specification | ||
|---|---|---|
| Task | Web code generation | Multimodal reasoning and coding |
| Published capabilities | Fine-tuned from Qwen3.5-4B-Base to generate a standalone page or slide deck from a brief. | Reuses its middle layers three times per token. Thinking Machines reports 48.2% on DeepSWE v1.1. |
| Model size | 4B class | 26.4B total / 3.4B active parameters |
| Access | Public 4.48 GB Q8_0 GGUF preview checkpoint | Downloadable weights and Tinker inference and customization |
| Result | A self-contained React HTML document | Not published |
| Operating limits | The publisher reports approximately 20,000 generated tokens for a typical whole document, including reasoning. | Native mixed-precision checkpoint occupies approximately 20 GB. |
| Sources | Model weights | Model weights · Additional documentation |
| Inputs | Text | Text, Images |
| Weights | Open weights · Apache-2.0 | Open weights · Apache-2.0 |
| Released | Nov 20, 2026 | Nov 20, 2026 |