Skip to content

GLM 5.2 Now Routes Direct To Z.ai — Free This Weekend

July 6, 2026

GLM 5.2 now connects straight to Z.ai's API instead of going through OpenRouter — lower latency, the same 1M context and agentic capabilities, and the same premium-band rates ($1.40 in / $4.40 out per 1M tokens). The old GLM 5.2 (via OpenRouter) entry is retired from the picker. This weekend onlyFriday 10 July through Sunday 12 July (UTC) — GLM 5.2 is completely free: no token-budget charge and no usage cost. Normal 6× premium billing resumes automatically on Monday 13 July (UTC).

What you can do

  • Pick GLM 5.2 from the model selector — one entry, direct Z.ai route, no OpenRouter hop.
  • Run long-horizon agent and coding workflows with 1M-token context, reasoning, tool support, and swarm eligibility — unchanged from the June launch.
  • Use GLM 5.2 free on the promo weekend10–12 July 2026 (UTC) — on any plan; Auto can still route to it in the SOTA tier during the promo.
  • Keep the same premium pricing after the promo6× token multiplier and standard per-token rates from 13 July (UTC) onward.

Where this shows up

  • You want GLM 5.2 for a heavy engineering session this weekend. Switch to GLM 5.2 (or leave Auto on a plan that can escalate to SOTA) and run the full workflow without burning premium tokens or usage credits — through Sunday night UTC.
  • You pinned the old OpenRouter GLM 5.2 path. Nothing to do — your selection now auto-switches to the direct Z.ai route, so you get the faster routing and the free weekend automatically. Same model, same capabilities, no re-selection.
  • You're comparing latency on agentic coding. Direct Z.ai should feel snappier on tool-heavy turns because requests no longer pass through an extra OpenRouter relay.

Try it

  • "Switch to GLM 5.2 and refactor this module end-to-end — read the whole repo context, plan, then implement."
  • "Use GLM 5.2 with tools to audit this codebase for security issues and produce a prioritized fix list."
  • "Leave Auto on and ask for a multi-step research + deck build — if it picks GLM 5.2 this weekend, the run is free."

Heads up

  • Promo window is UTC. Free usage applies on 2026-07-10, 2026-07-11, and 2026-07-12 (UTC calendar dates). Convert to your local timezone — the free window ends at Monday 00:00 UTC.
  • Automatic revert. No action needed after the weekend — billing returns to the premium band (6× tokens) on 13 July (UTC).
  • OpenRouter GLM 5.2 retired. The separate GLM 5.2 (via OpenRouter) picker entry is inactive. GLM 5.2 Turbo (OpenRouter) remains available as a faster, cheaper GLM option.
  • Existing pins migrate automatically. Anyone who had GLM 5.2 (via OpenRouter) selected is transparently routed to the direct Z.ai entry — for routing, pricing, and the weekend promo. No action needed, and nothing keeps hitting OpenRouter.
  • Supersedes routing only. The June 17 launch note still describes GLM 5.2's capabilities; this note covers the direct Z.ai route, OpenRouter retirement, and weekend promo.

Built for the Alfrada platform.