webAI open-sources 8.8B TwIL-8B hitting 93% of frontier coding performance
dr_cintas · x · 2026-11-05
webAI has open-sourced TwIL-8B, an open-weights model of just 8.8B parameters (weights on Hugging Face) that claims near-frontier scores on several hard benchmarks: 93% of frontier performance on coding, 92% on instruction following, and 90% on contest math — small enough to run on local hardware.
David Stout argues the industry's years-long focus on maximum capability regardless of cost is shifting: the next era of AI will be defined by the cost of completing work, not raw intelligence, and small models catching up fast supports that thesis.
More from Infra
- Hugging Face shares its model optimization pipeline: profile, torch.compile, ship open kernels — ariG23498 · 2026-11-05
- Reddit proposal: use agents to digest fragmented custom inference engines into reusable building blocks — nasone32 · 2026-11-05
- LZ4 v1.11.0 ships multithreaded compression, hits 7,550 MB/s with 16 workers — jedisct1 · 2026-11-05
- Engineer praises turbopuffer as community 'sandbox tier list' makes rounds — davidcrawshaw · 2026-11-05
- Indie pretraining run hits 10T tokens of data, locks in Apple M5 and DGX Spark support — jon_durbin · 2026-11-05
- HunyuanVideo 1.5 T2V FP8 outputs noisy images on AMD RX 7900 XTX — Fragrant-Throat7048 · 2026-11-05