Zhipu Shatters Global AI Speed Record with GLM-5.1 High-Speed Version
China's renowned AI team Zhipu officially announced today the launch of a new GLM-5.1 Highspeed API for select enterprise clients. Codenamed "GLM-5.1-highspeed," this model has stunned the industry since its release, delivering an output speed of 400 tokens per second.
This figure directly surpasses the current global API speed benchmark set by major large model providers, showcasing strong technical dominance. Previously, the AI industry assumed that model speed and size were mutually exclusive, with higher speeds typically requiring compromises in model capabilities.
Breaking Industry Norms with Flagship Performance
The GLM-5.1 Highspeed version, however, shattered the long-held industry belief that "fast means small." For the first time in domestic large models, it successfully brings flagship-level technical capabilities and ultra-low latency into real-world production environments.
According to reports, this model was developed jointly by Zhipu's GLM team and the TileRT team. Both teams conducted deep, comprehensive system-level optimizations across three layers: the inference engine, the scheduling system, and the underlying infrastructure, moving away from traditional dynamic scheduling.
Three-Level Optimization Ensures Stable Output
On the technical front, the development team not only rewrote the core inference path of the model architecture to boost single-card throughput, but also reduced latency in high-concurrency scenarios through techniques such as dynamic batching. Meanwhile, collaborative optimizations around the infrastructure ensured that 400 TPS became a stable and usable production-level capability.
This high-speed model has broad application prospects, particularly suited for scenarios with strict response latency requirements. Whether for AI programming, real-time voice interaction, or frequent business decisions, the model is now available on Zhipu's MaaS platform for select enterprises.
Related article
ChatGPT Deep Integration Unlocks 300,000 Movies and TV Shows via Conversation
As streaming services saturate the market with endless content, helping users cut through the noise to find what they actually want has emerged as a critical industry hurdle.On April 9, major streaming providers officially announced the integration o
How to fix Core Web Vitals for better SEO ranking
Transform your video editing workflow with AI-powered transcriptsTable of Contents:IntroductionKey Advantages of Automated TranscriptionEnsuring Accuracy and Refining OutputLeveraging Search Capabilities in TranscriptsSeamless Import of Existing File
DingTalk Rebuilds Architecture With AliAI’s MuleRun, Launching Full-Agent Era
DingTalk’s new CEO and head of the Wukong Business Unit, Chen Yusen, revealed a major organizational restructuring in an internal memo issued this afternoon. By creating the Core Platform Business Department and fully merging the AI Agent team, DingT
Related Special Topic Recommendations
Comments (0)
0/500
China's renowned AI team Zhipu officially announced today the launch of a new GLM-5.1 Highspeed API for select enterprise clients. Codenamed "GLM-5.1-highspeed," this model has stunned the industry since its release, delivering an output speed of 400 tokens per second.
This figure directly surpasses the current global API speed benchmark set by major large model providers, showcasing strong technical dominance. Previously, the AI industry assumed that model speed and size were mutually exclusive, with higher speeds typically requiring compromises in model capabilities.
Breaking Industry Norms with Flagship Performance
The GLM-5.1 Highspeed version, however, shattered the long-held industry belief that "fast means small." For the first time in domestic large models, it successfully brings flagship-level technical capabilities and ultra-low latency into real-world production environments.
According to reports, this model was developed jointly by Zhipu's GLM team and the TileRT team. Both teams conducted deep, comprehensive system-level optimizations across three layers: the inference engine, the scheduling system, and the underlying infrastructure, moving away from traditional dynamic scheduling.
Three-Level Optimization Ensures Stable Output
On the technical front, the development team not only rewrote the core inference path of the model architecture to boost single-card throughput, but also reduced latency in high-concurrency scenarios through techniques such as dynamic batching. Meanwhile, collaborative optimizations around the infrastructure ensured that 400 TPS became a stable and usable production-level capability.
This high-speed model has broad application prospects, particularly suited for scenarios with strict response latency requirements. Whether for AI programming, real-time voice interaction, or frequent business decisions, the model is now available on Zhipu's MaaS platform for select enterprises.
ChatGPT Deep Integration Unlocks 300,000 Movies and TV Shows via Conversation
As streaming services saturate the market with endless content, helping users cut through the noise to find what they actually want has emerged as a critical industry hurdle.On April 9, major streaming providers officially announced the integration o
How to fix Core Web Vitals for better SEO ranking
Transform your video editing workflow with AI-powered transcriptsTable of Contents:IntroductionKey Advantages of Automated TranscriptionEnsuring Accuracy and Refining OutputLeveraging Search Capabilities in TranscriptsSeamless Import of Existing File
DingTalk Rebuilds Architecture With AliAI’s MuleRun, Launching Full-Agent Era
DingTalk’s new CEO and head of the Wukong Business Unit, Chen Yusen, revealed a major organizational restructuring in an internal memo issued this afternoon. By creating the Core Platform Business Department and fully merging the AI Agent team, DingT





Home






