GLM-5.2

Pricing — 1 offering(s)

Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.

Capability profile

Operator guidance

Zhipu AI's current flagship, successor to GLM-5.1 (which this update marks `generation: previous`). Route here for open-weights agentic coding at long context; it is the strongest-benchmarked (vendor-reported) Chinese-lab open coding model in this registry. Against DeepSeek V4-Flash — the other MIT 1M-context flagship in the same batch — GLM-5.2 is pricier but carries more coding-benchmark signal. Independent capability data beyond coding is not yet available.

Use cases

• Long-horizon agentic coding over large codebases needing up to 1M tokens of context
• MCP-integrated agent workflows (Zhipu's stated design target)
• Self-hosted deployment under MIT where a strong open coding model with very long context is wanted

Limitations

• All cited benchmarks are Z.ai's own — Zhipu shipped GLM-5.2 without an official benchmark suite, so no independent third-party reproduction exists
• Expert-routing / active-parameter figures are from independent coverage, not Zhipu's own materials (which confirm only the total-parameter scale)
• Instruction-following, multilingual, speed, and cost-efficiency have no published data — rated `unknown`
• Launch date differs by a few days across sources (2026-06-13 coding-plan vs 2026-06-17 HF publication) — month-only `released` used

Citations

• GLM-5.2 overview — Z.AI Developer Document
• GLM pricing — Z.AI Developer Document
• zai-org/GLM-5.2 — Hugging Face