Doubao-pro

Training compute

2.5×10²⁵ FLOP

Parameters

500B

Published

Oct 28, 2024

Doubao-pro is an AI model developed by ByteDance (China), first published in October 2024. It works in the language domain, on tasks such as language modeling/generation, question answering, text summarization and text classification.

Training it took an estimated 2.5×10²⁵ FLOP of compute (estimation method: operation counting). The model has 500,000,000,000 parameters. It was trained on roughly 8.3T datapoints.

Access: API access. Its weights are not openly released. Epoch AI rates the confidence of this record as likely.

Full record

Organization: ByteDance
Country of organization: China
Domain: Language
Task: Language modeling/generation, Question answering, Text summarization, Text classification
Training compute: 2.5×10²⁵ FLOP
Compute estimation method: Operation counting
Parameters: 500,000,000,000
Dataset size: 8.3T
Model accessibility: API access
Open weights: No
Epoch confidence: Likely