Skip to content
TheChatArchiveThe Chat Archive
Geminigemini-2.5-pro-exp-03-25-grounding· GoogleHugging Face LMArena Search Arena 24K

What is the most recent update from OpenAI

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Geminigemini-2.5-pro-exp-03-25-grounding’s answer.
  1. User

    What is the most recent update from OpenAI

  2. Gemini Assistant

    Based on the latest information, here are the most recent updates from OpenAI as of April 18, 2025:

    New Reasoning Models: o3 and o4-mini (Released April 16, 2025)[1][2]

    • OpenAI released new reasoning models, o3 and o4-mini, described as their "smartest and most capable models to date" [1][2].
    • These models are designed to "think" longer before responding and excel at tasks requiring complex reasoning, coding, math, and visual analysis .
    • A key feature is their ability to agentically use and combine all available tools within ChatGPT (like web browsing, data analysis, image generation) to fulfill user requests [3].
    • They are the first OpenAI models that can integrate uploaded images into their reasoning process, analyzing visuals like charts, graphics, photos, and even hand-drawn sketches .
    • o3 is positioned as the flagship reasoning model for complex queries, while o4-mini is optimized for speed and cost-efficiency [1].
    • These models are now available for ChatGPT Plus, Pro, and Team users[1][2][3]. Free users can also access o4-mini with rate limits [2]. ChatGPT Enterprise and Edu users are expected to get access soon [1].

    New API Models: GPT-4.1 Family (Released April 14, 2025)[4][5][6]

    • OpenAI launched a new family of models for developers via the API: GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano [4][5].
    • These models show significant improvements over GPT-4o, particularly in coding, instruction following, and understanding long contexts (up to 1 million tokens) [4].
    • GPT-4.1 achieves state-of-the-art results on coding benchmarks like SWE-bench Verified and multimodal long-context benchmarks like Video-MME [4].
    • These models feature a refreshed knowledge cutoff date of June 2024 [4].
    • GPT-4.1 models are currently only available through the API, though improvements are being gradually incorporated into GPT-4o within the main ChatGPT interface [4].
    • OpenAI plans to deprecate the GPT-4.5 Preview model in the API by July 14, 2025, encouraging developers to transition to GPT-4.1 [4].

    Other Recent Developments:

    • Codex CLI: OpenAI released an open-source coding agent called Codex CLI that runs from a user's terminal, designed to leverage the capabilities of the new o3 and o4-mini models .
    • Memory Upgrade: OpenAI is enhancing ChatGPT's Memory feature, allowing it to reference all past interactions for increased personalization (currently rolling out to Pro subscribers) [7].
    • Audio Models: New audio models powered by GPT-4o (gpt-4o-transcribe, gpt-4o-mini-transcribe, gpt-4o-mini-tts) have been released for speech-to-text and text-to-speech tasks [8].
    • GPT-4o Mini: This model, announced previously, is now available for deployment in Azure OpenAI Service, offering a balance of intelligence, speed, and cost-effectiveness with a 128K context window [8].

    In summary, OpenAI's most recent major updates involve the release of new reasoning models (o3, o4-mini) enhancing ChatGPT's capabilities, particularly in tool usage and image reasoning, and the launch of the improved GPT-4.1 model family for developers via the API.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.