{"slug": "openai-reportedly-completes-pretraining-run-bel-with-over-10-trillion-parameters", "title": "OpenAI reportedly completes pretraining run ‘Bel’ with over 10 trillion parameters", "summary": "OpenAI has reportedly completed a massive new pretraining run codenamed 'Bel' with over 10 trillion parameters, according to an unverified post on X by user @synthwavedd on August 25, 2026. If accurate, it would be OpenAI's first successful full-scale frontier pretraining since GPT-4o shipped in May 2024, and could serve as a foundation for future systems including 'Astra' and successors to GPT-6. OpenAI has not officially confirmed the existence of Bel.", "body_md": "Via openai.com\n\n# OpenAI reportedly completes pretraining run ‘Bel’ with over 10 trillion parameters\n\nThe rumored model would represent OpenAI's first major pretraining milestone since GPT-4o and could serve as a foundation for future AGI-capable systems\n\nOpenAI has reportedly completed a massive new pretraining run codenamed “Bel,” featuring over 10 trillion parameters. If accurate, it would mark the company’s first successful full-scale frontier pretraining since GPT-4o shipped in May 2024.\n\nThe claim originated from a post on X by user @synthwavedd on August 25, 2026. OpenAI has not officially confirmed the existence of Bel or released any technical specifications.\n\n## What we know about Bel\n\nAccording to the unverified reports, Bel follows a prior pretraining run codenamed “Doug” and is intended to serve as a base model for future systems. Those future systems reportedly include something called Astra, as well as successors to GPT-6, with post-training work expected to unlock capabilities that push closer to what the industry calls Artificial General Intelligence.\n\nOpenAI has reportedly struggled with scaling since GPT-4o launched. Multiple sources within the AI research community have pointed to interrupted training runs, safety-related pauses in reinforcement learning processes, and technical difficulties that stalled progress on larger models. The company’s compute scaling efforts have been discussed under the umbrella of Project Stargate.\n\n## Why the gap since GPT-4o matters\n\nThe period between May 2024 and now has been an unusually long stretch for OpenAI to go without shipping a new frontier base model. The company released incremental updates and fine-tuned variants, but the core pretraining hadn’t produced a confirmed successor at a meaningfully larger scale.\n\nDuring that same window, competitors didn’t sit still. Anthropic pushed forward with its Claude model family, Google DeepMind continued scaling Gemini, and a growing roster of open-source efforts from Meta and others kept closing the gap.\n\n## The AGI question and competitive stakes\n\nThe framing around Bel is explicitly tied to AGI ambitions. The reported pipeline—from Bel to Astra to whatever follows GPT-6—suggests OpenAI is thinking in terms of a multi-stage roadmap where each layer of post-training and fine-tuning extracts progressively more capable behavior from the base model.\n\nOn the other hand, the cautionary note is straightforward: none of this is confirmed. Until OpenAI provides official validation, whether through a technical paper, a product announcement, or even a casual blog post, Bel remains an unverified claim from a social media post.\n\n**Disclosure:** This article was edited by Editorial Team. For more information on how we create and review content, see our\n\n[Editorial Policy](https://cryptobriefing.com/editorial-policy/).", "url": "https://wpnews.pro/news/openai-reportedly-completes-pretraining-run-bel-with-over-10-trillion-parameters", "canonical_source": "https://cryptobriefing.com/openai-bel-pretraining-10-trillion-parameters/", "published_at": "2026-08-25 19:13:57+00:00", "updated_at": "2026-08-25 19:45:00.733243+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-research"], "entities": ["OpenAI", "Bel", "GPT-4o", "Astra", "GPT-6", "Anthropic", "Google DeepMind", "Meta"], "alternates": {"html": "https://wpnews.pro/news/openai-reportedly-completes-pretraining-run-bel-with-over-10-trillion-parameters", "markdown": "https://wpnews.pro/news/openai-reportedly-completes-pretraining-run-bel-with-over-10-trillion-parameters.md", "text": "https://wpnews.pro/news/openai-reportedly-completes-pretraining-run-bel-with-over-10-trillion-parameters.txt", "jsonld": "https://wpnews.pro/news/openai-reportedly-completes-pretraining-run-bel-with-over-10-trillion-parameters.jsonld"}}