OpenAI在Microsoft Foundry推出GPT-chat-latest模型

robot
Abstract generation in progress

AIMPACT News, May 6th (UTC+8), OpenAI recently launched the latest chat model GPT-chat-latest (corresponding to GPT-5.5 Instant in the API) at Microsoft Foundry. This model is built on GPT-5.4 and GPT-5.3-chat, achieving measurable improvements in factual accuracy, tool invocation, and response efficiency. Compared to GPT-5.3-chat, hallucination frequency is reduced by 52.5%, hallucination statements by 37.3%; it performs excellently across multiple benchmarks: scientific chart reasoning increased from 75.0 to 81.6, expert multimodal reasoning from 69.2 to 76.0, PhD-level scientific questions from 78.5 to 85.6, and competitive math from 65.4 to 81.2. In terms of output efficiency, the number of words in common prompts is reduced by approximately 25-30%, and tool invocation and retrieval capabilities are also improved. The model aims to support practical workflows such as multi-turn dialogue assistants, agent systems, and retrieval-augmented applications. (Source: InFoQ)

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pin