webAI releases TwIL-8B reasoning model, 98.2 on MATH-500

webAI released TwIL-8B, an 8.8B-parameter reasoning model post-trained from IBM's Granite 4.2 8B. It scores 98.2 on MATH-500 and 81.6 on LiveCodeBench but trails similar-size models on knowledge benchmarks.

Why it matters: Near-saturated math scores from an 8B model show reasoning post-training can be bolted onto small enterprise base models cheaply. The weak knowledge results are the usual trade-off: reasoning gains do not add facts, so such models suit tool-augmented use more than open-ended Q&A.