OpenClaw Unveils Major Update with GPT-5.4 Support and Memory Hot Swapping
The highly anticipated open-source agent project OpenClaw has launched its core version 2026.3.7. This release introduces powerful new capabilities like the Context Engine plugin and distributed channel binding, while also adding full support for OpenAI's latest flagship model, GPT-5.4.

OpenClaw has now surpassed 280,000 stars on GitHub. The development team notes that the project is evolving from an experimental agent framework into a production-ready "agent operating system" designed for high persistence and scalability.
Key highlights of this update include:
Context Engine Plugin: A new Context Engine interface allows developers to integrate custom RAG (Retrieval-Augmented Generation) or lossless compression algorithms. This addresses the "forgetfulness" problem in long agent conversations, enabling reliable long-term autonomous operation.
Native Support for GPT-5.4 and Gemini: The update integrates GPT-5.4, which demonstrated superior performance compared to official native tools in the OOLONG benchmark. It also includes deep optimizations for the Google Gemini 3.1 series models.
Persistent Channel Binding: Using the ACP protocol, agents can form deep, persistent connections with channels on Discord or Telegram. These connections survive server restarts, allowing agents to "remember" previous conversation context and progress for truly continuous service.
Local Model Experience Optimization: The streaming output for local models like Ollama has been restructured. This change cleanly separates the model's internal "reasoning process" from its final response, preventing intermediate reasoning drafts from disrupting the user experience.
In performance tests, the new version of OpenClaw, combined with the lossless-claw plugin, delivered impressive results when processing extremely long contexts. It achieved a score of 74.8 in long-text comprehension tasks, outperforming competing solutions.
As OpenClaw's popularity grows, its applications are expanding beyond programming into areas like stock analysis, report generation, and office automation. A growing number of users are deploying it as a cost-effective "virtual employee" to handle routine daily tasks.
Related article
Apple Smart Glasses Could Debut at WWDC27, Highlighting Privacy Protection
Bloomberg’s Mark Gurman reports that Apple’s smart glasses, codenamed N50, are slated for a WWDC27 debut in June 2027, with a retail launch expected in autumn 2027. Originally targeted for late this year and early 2027, the device’s release has been
Inside Details Exposed About Next-Gen Gemini: Strained Computing Power, Internal Teams Disagreed on Development Priorities and Resource Allocation
Reports indicate that the launch of Google’s highly anticipated next-generation Gemini model has been pushed back. Internal disagreements over development priorities and resource allocation, combined with limited computing capacity and complex approv
OpenAI Dismisses Growth Slowdown Concerns, Says Multiple Business Units Accelerating
In response to external scrutiny regarding decelerating sales growth and missed internal benchmarks, AI leader OpenAI issued a confident statement on Tuesday, April 28. The company clarified that its consumer products and enterprise services are adva
Related Special Topic Recommendations
Comments (0)
0/500
The highly anticipated open-source agent project OpenClaw has launched its core version 2026.3.7. This release introduces powerful new capabilities like the Context Engine plugin and distributed channel binding, while also adding full support for OpenAI's latest flagship model, GPT-5.4.

OpenClaw has now surpassed 280,000 stars on GitHub. The development team notes that the project is evolving from an experimental agent framework into a production-ready "agent operating system" designed for high persistence and scalability.
Key highlights of this update include:
Context Engine Plugin: A new Context Engine interface allows developers to integrate custom RAG (Retrieval-Augmented Generation) or lossless compression algorithms. This addresses the "forgetfulness" problem in long agent conversations, enabling reliable long-term autonomous operation.
Native Support for GPT-5.4 and Gemini: The update integrates GPT-5.4, which demonstrated superior performance compared to official native tools in the OOLONG benchmark. It also includes deep optimizations for the Google Gemini 3.1 series models.
Persistent Channel Binding: Using the ACP protocol, agents can form deep, persistent connections with channels on Discord or Telegram. These connections survive server restarts, allowing agents to "remember" previous conversation context and progress for truly continuous service.
Local Model Experience Optimization: The streaming output for local models like Ollama has been restructured. This change cleanly separates the model's internal "reasoning process" from its final response, preventing intermediate reasoning drafts from disrupting the user experience.
In performance tests, the new version of OpenClaw, combined with the lossless-claw plugin, delivered impressive results when processing extremely long contexts. It achieved a score of 74.8 in long-text comprehension tasks, outperforming competing solutions.
As OpenClaw's popularity grows, its applications are expanding beyond programming into areas like stock analysis, report generation, and office automation. A growing number of users are deploying it as a cost-effective "virtual employee" to handle routine daily tasks.
Apple Smart Glasses Could Debut at WWDC27, Highlighting Privacy Protection
Bloomberg’s Mark Gurman reports that Apple’s smart glasses, codenamed N50, are slated for a WWDC27 debut in June 2027, with a retail launch expected in autumn 2027. Originally targeted for late this year and early 2027, the device’s release has been
Inside Details Exposed About Next-Gen Gemini: Strained Computing Power, Internal Teams Disagreed on Development Priorities and Resource Allocation
Reports indicate that the launch of Google’s highly anticipated next-generation Gemini model has been pushed back. Internal disagreements over development priorities and resource allocation, combined with limited computing capacity and complex approv
OpenAI Dismisses Growth Slowdown Concerns, Says Multiple Business Units Accelerating
In response to external scrutiny regarding decelerating sales growth and missed internal benchmarks, AI leader OpenAI issued a confident statement on Tuesday, April 28. The company clarified that its consumer products and enterprise services are adva





Home






