XNews

GLM-5.3 Model Launch and Benchmark Performance

GLM-5.3 Model Launch and Benchmark Performance

Zhipu AI launched the GLM-5.3 API on Aug 18, 2026. The model is built for coding, defensive cybersecurity, and long-horizon agentic tasks. It is priced the same as GLM-5.2 and available via the official API and partner model gateways. On Terminal-Bench, GLM-5.3 released and beat Fable and the new DeepSeek V4-Pro 0813 from the previous day. GLM-5.3 is the newest member of the cost-accuracy pareto for Terminal-Bench 3.0, replacing Grok 4.6. On the Artificial Analysis Intelligence Index, GLM-5.3 achieved 60, on par with Kimi K3 and up 7 points from GLM-5.2. In Design Arena, GLM-5.3 placed 3rd overall with an Elo of 1351, a 6-position improvement.

Key insights

GLM-5.3 API launches with unchanged pricing

Zhipu AI made the GLM-5.3 API live. The model is built for coding, defensive cybersecurity, and long-horizon agentic tasks. It is priced the same as GLM-5.2. Availability is through the official API and partner model gateways. The launch included a getting-started link for access.

Terminal-Bench performance and cost-accuracy position

GLM-5.3 released and beat Fable and DeepSeek V4-Pro 0813, which had appeared the previous day, on Terminal-Bench. The model is described as the newest member of the cost-accuracy pareto for Terminal-Bench 3.0, replacing Grok 4.6.

Artificial Analysis and Design Arena rankings

GLM-5.3 achieved 60 on the Artificial Analysis Intelligence Index, on par with Kimi K3 and up 7 points from GLM-5.2. In Design Arena, GLM-5.3 placed 3rd overall with an Elo of 1351, which represented a 6-position improvement.

Source posts