webAI's 3.6B TwIL-LM3-Pro beats VibeThinker-3B by 35% on formal logic, runs locally in 2GiB
rohanpaul_ai · x · 2026-10-01
webAI released TwIL-LM3-Pro, a 3.66B-parameter open model built for local reasoning: the recommended Q4 build is a 2.09GiB file that runs via llama.cpp on CPU or local GPU, keeping private data on-device. It hit 500K downloads in its first month.
Key claims from webAI's evaluation:
- Highest recorded formal-logic score among small models compared, beating Weibo's VibeThinker-3B by 35%, Qwen3.5-4B by 24%, and Liquid AI's LFM2.5-8B-A1B by 47%
- SVAMP 95% and MuSR 64.1%, both tops among compared small models
- Built on IBM Granite with a 28% lift in formal-logic scoring
More from Models
- Mystery Model 'Gestalt' Tops New Frontier Design Benchmark by 214 Points — soleio · 2026-11-19
- Gemini 4 Neon arrives days after Hydrogen: fast reasoning at $3/$15 per 1M tokens — TorturedPoet30 · 2026-11-19
- Screenshot confirms Gemini 4 Neon launch at $3/$15 per million tokens — Recoil42 · 2026-11-19
- mrfakename releases abliterated GLM-5.3-Flash and Qwen3.8-Flash-Next — mrfakename0 · 2026-11-19
- Physician: AI competition now measured in weeks, GPT-6 versions 7 days apart — ramos_casals · 2026-11-19
- micro1's flow qc model screens training data for benchmark leaks before release — Exp_Mark · 2026-11-19