--- format: "aidr-story-markdown/v1" id: "f08685da9fdd642e9e0aea16eecdf6c1f3ac565398b2e2447a1de405cd790068" canonical_url: "https://aidr.today/f08685da?lang=en" title: "webAI Open-Sources TwIL-8B AI Model to Rival Frontier Performance Benchmarks" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-11-04T19:39:00.000Z" category: "Open Source" topics: ["open-source","llm"] source_urls: ["https://marketbrief.now/ai/webai-open-sources-twil-8b-ai-model-to-rival-frontier-performance-benchm-c47b77ae","https://huggingnews.com/ai/webai-open-sources-twil-8b-ai-model-to-rival-frontier-performance-benchm-c47b77ae","https://news.ycombinator.com/item?id=50387546","https://village.dev/blog/closed-ai-models-now-dominate-the-price-performance-frontier/"] summary: "webAI open-sources the TwIL-8B AI model, which utilizes 8.8B parameters. The model scored 93% on coding, 92% on instruction following, and 90% on contest mathematics. This compact model is designed to run on personal hardware as an alternative to expensive API calls." --- # webAI Open\-Sources TwIL\-8B AI Model to Rival Frontier Performance Benchmarks > [Open the canonical story]() **Published:** 2026-11-04T19:39:00.000Z **Category:** Open Source **Topics:** open\-source, llm ## Summary webAI open\-sources the TwIL\-8B AI model, which utilizes 8\.8B parameters\. The model scored 93% on coding, 92% on instruction following, and 90% on contest mathematics\. This compact model is designed to run on personal hardware as an alternative to expensive API calls\. ## Sources - [Story source]() - [Story source]() - [Discussion]() - [Story source]()