Expired50% confidenceEventExact time
Reddit community leaker Leo claimed that OpenAI has just completed its next pre-trained model exceeding 10 trillion parameters in scale, codenamed 'Bel'.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
9/12/2026
First Seen
Valid until: 9/26/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedOpenAI has publicly discussed that their model was specifically trained to efficiently use the apply_patch tool.68% similarVerifiedOpenAI's o1 model was released in September 2024; through reinforcement learning it was trained to perform internal reasoning before outputting a final answer, with intermediate reasoning steps existing as special tokens that are invisible to users.68% similarUnverifiedWhen training dexterous hand manipulation of a Rubik's Cube, OpenAI used thousands of parallel simulation instances to collect experience data simultaneously.67% similarUnverifiedOpenAI publicly warned that frontier models already possess the capability to launch sophisticated cyberattacks, and suspended training of some frontier models.67% similarPartially VerifiedOpenAI internally once used three people over five to six months to generate more than one million lines of code through Harness Engineering and successfully launched it.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/913785API
curl https://kongchang.com/api/v1/knowledge/claims/913785MCP
get_claim(id=913785)