Thursday, November 5, 2026 Today’s Paper
Log in Join free
Vol. I · No. 37 Texas Edition
Thursday, Nov 5
Xultr
Texas first. Then the world.
Always free Updated just now
Read. React. Discuss.
Tech

webAI Releases TwIL-8B, an 8.8B Reasoning Model Scoring 98.2 on MATH-500

MarkTechPost By Asif Razzaq 4 min read 0 views 0 comments
webAI Releases TwIL-8B, an 8.8B Reasoning Model Scoring 98.2 on MATH-500
Image: MarkTechPost

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

The post webAI Releases TwIL-8B, an 8.8B Reasoning Model Scoring 98.2 on MATH-500 appeared first on MarkTechPost.

MarkTechPost Original story · www.marktechpost.com
Read the full story
How does this story make you feel?
👍 ❤️ 💡 😮 😂 😢 😠

Discussion (0)

No comments yet. Start the conversation!

More to read