GLM-5.1 is the next generation flagship model for agent engineering, using a hybrid expert (MoE) architecture and has 754B parameters. The model's programming capabilities have been significantly enhanced, achieving leading levels on SWE-Bench Pro, and significantly leading the previous generation on benchmarks such as NL2Repo (warehouse generation) and Terminal-Bench 2.0 (real terminal tasks). GLM-5.1 is specially designed for long-term agent tasks...
Provider: Z.ai
Context window: 198K
Pricing: $0.67 input / $2.21 output per 1M tokens
Capabilities: serverless, json_mode, function_calling, structured_output
