modelgrep

DeepSeek: DeepSeek V3.2 vs Z.ai: GLM 4.7

Z.ai: GLM 4.7 wins on more metrics (6 of 9), but the right pick depends on what you optimize for — see the breakdown below.

MetricDeepSeek: DeepSeek V3.2Z.ai: GLM 4.7
Intelligence Index41.742.1
Coding Index36.736.3
GPQA Diamond84%86%
Design Arena Elo12211274
Speed (tokens/sec)41179
Latency458ms328ms
Input price /M$0.229$0.400
Output price /M$0.343$1.75
Context window131K203K
CapabilitiesReasoningToolsJSONReasoningToolsJSON