GLM 5.2 — Specifications
| Developer | Zhipu AI |
|---|
| Type | LLM (coding/agentic, MoE) |
|---|
| Modality | Text → Text |
|---|
| Parameters | 744B total / ~40B active (MoE) |
|---|
| Context window | 1M |
|---|
| Max output | 131K |
|---|
| License | MIT (open) |
|---|
| Open weights | ✅ Yes |
|---|
| Released | 2026-06 |
|---|
| Input price | $1.4 /1M |
|---|
| Output price | $4.4 /1M |
|---|
| API providers | Zhipu (Z.ai), OpenRouter |
|---|
🖥️ Run it locally
| VRAM (4-bit) | ~370 GB |
|---|
| Minimum GPU | Multi-GPU server (e.g. 5× H100 80GB) |
|---|
Official page →
Zhipu AI’s open 1M-context model — a 744-billion-parameter mixture-of-experts (≈40B active) using a new ‘IndexShare’ sparse-attention design, notably trained entirely on Huawei chips. MIT open weights; strong on coding, design and agentic tasks.
