OpenAI Releases New /search API and Two Audio Models to Enhance Web Search and Intelligent Agent Development Capabilities.CEO Sam Altman Says Codex is a Significant Advancement for AGI at Snowflake Summit.Gemma Technology Enables Users to Deploy AI Agents on Local Hardware.New Codex Features Allow Internet Access.v0 Platform Enables Full-stack application development with integration of OAuth, OpenAI and Stripe. precision voice control technology OpenAudio S1 breaks through in TTS.
- This summary is generated by AI analyzing the content of the article for informational purposes only.
We are pleased to announce that the highly anticipated /search API is now live.
One of our most popular endpoints, the /search API is designed to provide developers and agents with an extremely easy way to discover and crawl web content. With a single API call, users can easily search the entire Internet and extract the results in a format suitable for Large Language Models (LLMs).
Key features include:
- Efficient Search: Utilize advanced algorithms to quickly retrieve relevant content.
- Seamless Integration: Supports multiple programming languages and development platforms.
- Data Formatting: Automatically converts captured data into a format that LLM can use directly.
- Real-time updates: Ensure access to the latest and most relevant information.
This feature is now available in our API, Management Control Panel (MCP), and all integrations.
Whether you need to analyze data on a large scale or want to simplify your daily development workflow, the /search API provides you with powerful support.
---
At the Snowflake Summit on June 2, 2025, OpenAI CEO Sam Altman shared his thoughts on the future of artificial intelligence.
Altman says that their recently launched coding agent, Codex, is one he's experienced firsthandGeneral Artificial Intelligence (AGI)At all times.Codex is capable of handling a wide range of tasks and working autonomously in the background with great performance. Not only is it able to accomplish complex tasks, but it will also be able to observe meetings, read Slack messages and internal documents in the near future.
"These models are going to be pretty amazing in the next year or two." Altman said."We have a lot of room for improvement and a lot of improvements coming."
He further emphasized that these models will be able to understand all the contextual information provided by the user, connect to a variety of tools and systems, and perform in-depth reasoning to ultimately give answers. They are robust enough for users to trust them to do the job independently.
Altman believes that this capability is very close, and that excites him.
[Full video: https://t.co/3HV0GvQ51r]
---
According to the latest news, OpenAI is gearing up for the June 3, 2025 release of two new models that support native audio processing:gpt-4o-audio-preview-2025-06-03 respond in singing gpt-4o-realtime-preview-2025-06-03. These two new models will significantly enhance multimodal processing, especially in the audio domain.
Among them.gpt-4o-audio-preview-2025-06-03 It aims to provide more powerful audio understanding and generation functions for application scenarios such as speech recognition, audio to text, and speech synthesis. While gpt-4o-realtime-preview-2025-06-03 On the other hand, it focuses on real-time audio processing, enabling real-time dialog, voice command understanding and instant feedback.
This series of updates will further advance the application of artificial intelligence in the field of audio processing, providing users with a richer and more natural interactive experience.
---
In the latest news, Claude Research and remote MCP integration is now officially available to Pro users.
The Claude Research platform provides a range of advanced natural language processing tools that can help researchers and developers perform model training and data analysis more efficiently. Remote MCP (Model Control Panel) integration allows users to monitor and adjust models in real-time via remote access, greatly enhancing development efficiency.
Key features include:
- Advanced Natural Language Processing Tools
- Real-time model monitoring and adjustment
- Flexible remote access capabilities
These new features will enable Pro users to better utilize the power of the Claude platform to accelerate project progress.
---
Recently, OpenAI released four important updates designed to further enhance its ability to build intelligent agents.
First, the **Agents SDK** now supports TypeScript, which will provide developers with more powerful type checking and a better development experience. Second, the new **RealtimeAgent** feature, designed for voice agents, enables real-time dialog processing, significantly enhancing the user experience.
In addition, **Traces** support has been added to the Realtime API, allowing developers to trace and record real-time data streams to better debug and optimize systems. Finally, OpenAI has made improvements to its **Speech Synthesis Model** to provide more natural and smooth speech output.
These updates not only enhance the flexibility and usability of the technology, but also provide developers with additional tools to build more efficient and intelligent agent applications.
---
A detailed overview of using Gemma to build custom AI agents on personal hardware. This technology has revolutionized the way users can create and run AI agents completely locally on their own laptops.
No need to connect or pay a provider like OpenAI or Anthropic. Combine the open source Google Gemma 3 model, Smolagents and LM Studio to build a complete local AI agent environment.
Key Strengths:
- **Fully localized**: all data and processing is done locally for increased privacy and security
- **No need to rely on cloud services**: avoiding dependency on third-party services reduces costs
- **High flexibility**: AI agent functionality and performance can be customized to meet specific needs
This is a huge boon for developers and researchers, especially in application scenarios that require a high degree of privacy protection, such as medical data analytics, financial transactions, and so on.
---
Recently, OpenAI announced the release of its programming assistant, Codex, to Plus users and introduced a new feature that enables the Codex Agent to access the Internet.
This new feature has gone live today and allows users to enable Internet access by creating a new environment or editing an existing one. By default, Internet access is turned off and users have full control over the domain names and HTTP methods that Codex can use when performing tasks.
Key Points:
- Internet access needs to be enabled manually
- Users can customize the domain name and HTTP method for Codex access
- Please refer to the official documentation for more details on how to use it and the risks involved.
This update provides developers with greater flexibility and functionality, enabling Codex to be used in a wider range of scenarios. For example, developers can use Codex to fetch real-time data from the Internet, perform dynamic code generation, or call external APIs, among other things.
For more detailed information and instructions, please visit:https://t.co/9LbzFI6WUb
---
The kithbook_com team is building full-stack applications using the v0 platform, integrating several advanced technologies. Currently, the team has completed the Google OAuth authentication feature and has 80 people on the waiting list, 4 of whom paid $3 for early access.
Next steps include enabling synchronization of Gmail and Google Calendar, integrating OpenAI, and building user profiles with context. These features will greatly enhance the user experience and provide users with a more personalized and intelligent service.
Major Updates:
- Google OAuth authentication has been successfully implemented
- 80 people have been added to the waiting list
- Four people paid $3 for early access to the
Next Steps Program:
- Synchronization of Gmail and Google Calendar
- Integration of OpenAI technologies
- Building user profiles with contextual information
With the v0 platform, the kithbook_com team was able to efficiently develop and deploy these features to provide users with a seamless and secure application experience.
---
Recently, a precise voice control technology called OpenAudio S1 has attracted widespread attention. This technology not only makes a breakthrough in the field of speech synthesis (TTS), but also provides users with an unprecedented interactive experience.
OpenAudio S1 achieves highly natural and expressive speech synthesis with advanced natural language processing. It was rated first in the TTS-Arena evaluation with a Word Error Rate (WER) of only 0.008 and a Character Error Rate (CER) of 0.004, which is a leading position in the Seed TTS evaluation.
The core features of OpenAudio S1 include:
- Unparalleled expressiveness and naturalness: the ability to accurately convey a wide range of emotions, from anger and happiness to sadness to subtle emphasis, whispers and empathy.
- Wide range of application scenarios: suitable for video production, audiobooks, podcasts, AI companions, games and many other fields.
The emergence of this technology has greatly enriched the possibilities of human-computer interaction and provided creators with more means to realize their creativity.
With the continuous progress of speech recognition and synthesis technology, precise voice control will become an important part of future smart devices, further promoting the development of artificial intelligence technology.
---
Competitive analysis has never been easier, thanks to the new /search endpoint from @firecrawl_dev. This innovative tool dramatically improves the efficiency of market research by automating the process.
The tool builds a robust pipeline system that automatically identifies competitors and crawls key information from their websites, finally generating detailed reports. The entire process is driven by @LangChainAI's StateGraph, ensuring smooth execution of the workflow.
**Main features include:**
- Automatic discovery and identification of competitors
- Efficient crawling of target website data
- Generate comprehensive competitive analysis reports
With these features, organizations can understand market dynamics faster and make more informed business decisions.
**Wide range of application scenarios:**
- Marketers can use this tool to conduct competitive analysis and optimize promotion strategies;
- Product managers can improve their own products by analyzing the features of competitors' products;
- Business decision makers can develop more competitive strategies based on detailed data reports.
In short, @firecrawl_dev's new /search endpoint provides strong support for competitive analysis and greatly improves productivity.
Comments are closed.