← Catalog
TwIL-8B
AvailablewebAIOpen weights
8.79B dense reasoning model post-trained from IBM Granite 4.2 8B with full-parameter SFT then GRPO/DAPO reinforcement learning, specialized for contest math, code and instruction following. Ships as balanced (default) and verbose variants that share one architecture and tokenizer; 131,072-token context; English text in/out. Lab-reported MATH-500 98.2 on the balanced variant. Released under the webAI Non-Commercial License ver. 1.0.
Specifications
- License
- Open weights · webAI Non-Commercial License ver. 1.0
- Weights
- Downloadable
- Architecture
- Dense
- Parameters
- 8.79B
- Context window
- 131K tokens
- Max output
- —
- Knowledge cutoff
- —
- Price (in / out, $/M)
- —
- Modalities
- Text
Benchmarks
No benchmark scores recorded yet. Spotted some? Submit a correction.
Vendor-reported figures are claims until independently verified. See methodology.