GLM-5.2
#25 in Frontier LLMszhipu-ai · v5.2 · since 2026-06-13 (GLM Coding Plan); Weights & API: 2026-06-16/17 · 14× · last seen Jul 03, 2026
GLM-5.2 is an open-weight large language model from Zhipu AI (brand: Z.ai), initially released on June 13, 2026 for GLM Coding Plan subscribers, with MIT-licensed weights on Hugging Face and a standalone API going live on June 16–17, 2026. It uses a Mixture-of-Experts architecture with 753 billion total parameters (~40 billion active) and a 1-million-token context window, kept inference-efficient by the new IndexShare sparse-attention technique. The model is purpose-built for long-horizon agentic coding and autonomous workflows, and ranks as the strongest open-weight model on SWE-bench Pro and Terminal-Bench 2.1 according to independent evaluations. Z.ai API pricing is $1.40/1M input tokens and $4.40/1M output tokens.
Features
| Key Benchmark (%) | SWE-bench Pro: 62.1% | Terminal-Bench 2.1: 81.0% | GPQA Diamond: 91.2% | Artificial Analysis Intelligence Index v4.1: 51 (Rank 6 open-weight) |
| Context Window (Tokens) | 1,000,000 tokens (1,048,576); max. output: 131,072 tokens (API); MoE 753B total / ~40B active |
| License | MIT (open weights, commercial use, fine-tuning, and self-hosting allowed) |
| Multimodality | Primarily text; vision input (image) supported per Requesty/API card; tool calling, reasoning (High/Max), structured output |
| Platform | Z.ai (zhipuai.cn) API; Hugging Face (zai-org/GLM-5.2); OpenRouter; Cloudflare Workers AI; ModelScope; self-hosting via vLLM, SGLang, KTransformers, Ollama |
| Price per 1M Tokens | $1.40 input / $4.40 output (Z.ai API); cache input: $0.26; GLM Coding Plan subscription: from $18/month |
| Release Date | June 13, 2026 (GLM Coding Plan); weights + standalone API: June 16/17, 2026 |