Comparison

GLM Max vs Google AI Ultra $200 (Antigravity) for coding

At p50 peak, GLM Max returns 0.27 building h/week per $/month (45 h/week for $168/mo) and Google AI Ultra $200 (Antigravity) returns 0.23 building h/week per $/month (45 h/week for $200/mo), so GLM Max gives more building hours per dollar; GLM Max runs a benchmark-measured model for building (54.0/100 quality), the other does not; Google AI Ultra $200 (Antigravity) hours are guessed — treat them as a guess.

GLM Max vs Google AI Ultra $200 (Antigravity), side by side

Same catalog, same clock: p50 with p10–p90 intervals. Quality is each seat's coding-agent benchmark score with its agent and reasoning setting.

GLM MaxGoogle AI Ultra $200 (Antigravity)
Price$168/mo$200/mo
Building h/week p50 (p10–p90), peak45 (24–90)45 (18–90)
Building h/week p50 (p10–p90), off-peak90 (48–180)—
Building h/week per $/mo at p50 peak0.270.23
Benchmark quality /100sGLM-5.3 · Opencode · 54/100 planning · GLM-5.3 · Opencode · 54/100 building · GLM-5.3 · Opencode · 54/100 adminnot measured planning · not measured building · not measured admin
Sourceofficial-clock + livedguessed
Terminal CLI agentYesYes
Clockpeak Mon–Fri 07:00–11:00 Londonlimits refresh every 5h, weekly cap; no off-peak rate