- GLM-5.1-highspeed reaches an output speed of 400 tokens/s, combining flagship capabilities with ultra-low latency for the first time among domestic large models.
- The release follows the open-sourcing and price hike of its flagship GLM-5.1 model.
Chinese artificial intelligence startup Zhipu has launched GLM-5.1-highspeed, an ultra-fast version of its flagship model. The new product achieves an output speed of 400 tokens per second (tokens/s), setting a new global record for API speeds among large model providers.
GLM-5.1-highspeed breaks the industry convention where fast models are typically lightweight, bringing flagship-level capabilities and ultra-low latency to production environments simultaneously for the first time, the company said on Friday.
Co-developed by Zhipu and the TileRT team, the model underwent system-level optimizations to its underlying architecture.
It abandons dynamic scheduling at the runtime layer, significantly boosting single-card throughput and reducing tail latency through static orchestration.
The high-speed API is currently available to select enterprise customers on the Zhipu MaaS (Model-as-a-Service) platform.
It is particularly suitable for application scenarios highly sensitive to response latency, such as AI coding, real-time interaction, and business decision-making, according to the announcement.
The launch follows Zhipu's move to open-source its GLM-5.1 model in April this year. At that time, the company raised the model's price by 10%, aligning its pricing in core scenarios with Anthropic's products.
Zhipu, which listed in Hong Kong earlier this year, reported that its total revenue for 2025 reached 724 million yuan ($106.6 million). This represents a 131.9% year-on-year growth, primarily driven by its cloud deployment business.
However, the increasingly fierce AI race and the continuous costs of model iteration have weighed on its profitability.
The company's net loss widened to 4.72 billion yuan in 2025, reflecting the massive investment requirements in the sector.
($1 = 6.7921 yuan)
