Z.ai's subscription for GLM models: GLM-5.3 flagship plus GLM-5.3-Flash (requests for older 5.2/5.1 auto-route to 5.3, and 4.7 to 5.3-Flash), metered by credit-based 5-hour and weekly windows, with bundled Vision and Web Search/Reader MCPs.
Pricing: Lite $18/mo (verified on official docs); Pro ~$80 and Max ~$168 per third-party trackers — confirm at checkout — with quarterly and yearly discounts. Credit windows: 2,000 / 12,000 / 28,000 per 5 hours and 10,000 / 60,000 / 140,000 per week (Lite/Pro/Max).
At $18 the GLM Coding Plan is one of the best cheap flat plans in the field — if GLM-5.3 alone covers your work, it's the value pick, and its Anthropic-compatible endpoint backs Claude Code, which we can't. The catches are single-vendor scope and credit windows whose burn varies by model. Standard Compute is the alternative when you need more than one model family or your volume outruns the windows: flat-rate across frontier and open models, with no credits to count.
Standard Compute is an OpenAI-compatible API with frontier-model compute at a flat monthly price (from $39/mo) — no per-token billing, no rate-limit windows. Under extreme sustained load requests are paced smoothly instead of erroring or charging more.
Standard Compute is OpenAI-compatible, so any tool or SDK that lets you set a custom base URL migrates in minutes:
Base URL = https://api.stdcmpt.com/v1 API key = your Standard Compute key Model = standardcompute
Setup guides for every major agent — OpenClaw, Hermes, OpenCode, Cursor, Cline, Aider and more — on the integrations page. Free tier to test it, no card required.
No — since the 2026 revamp it meters usage in credits across dual windows (per-5-hour and weekly, per tier). Credit burn varies by model, so 'how much do I have left?' takes arithmetic. Standard Compute has no credit meter; sustained heavy load is paced smoothly instead of counted down.
GLM Coding Plan — it has an Anthropic-compatible endpoint with official Claude Code docs, and Standard Compute can't power Claude Code at all. For OpenAI-compatible agents (OpenCode, Cline, Aider) the comparison is real: GLM's $18 wins if one model family inside credit windows fits; flat-rate multi-model wins if it doesn't.
The current flagship GLM-5.3 plus GLM-5.3-Flash. Requests for older models are auto-routed up (5.2/5.1 to 5.3, 4.7 to 5.3-Flash), so you're always on the newest generation — but only ever on GLM.
The flat-rate alternative to GLM Coding Plan (Z.ai)