OpenAI发射GPT-4o和GPT-4o小型微调支持开发者。 OpenAI launches GPT-4o and GPT-4o mini fine-tuning support for developers.

OpenAI公司为GPT-4o和GPT-4o小型公司启动了微调支持,使开发商能够根据定制数据集的具体使用案例调整AI模型。 OpenAI has launched fine-tuning support for GPT-4o and GPT-4o mini, enabling developers to adapt the AI models to specific use cases with custom datasets. 所有有偿使用等级都可用,其特点可以降低成本,改进性能。 Available to all paid usage tiers, the feature allows for lower costs and improved performance. 开发人员在9月23日之前,将获得1万个GPT-4o微调代币和2万个GPT-4o迷你代币,之后,培训费用为25美元/百万个代币,推断为每百万个输入/输出代币3.75美元/15美元. Until Sept 23, developers receive 1M tokens for GPT-4o fine-tuning and 2M for GPT-4o mini, after which, training costs $25/million tokens, with inference at $3.75/$15 per million input/output tokens.

August 20, 2024

19 文章