zai-org/GLM-5.2

GLM-5.2: Long-Horizon Task Specialist

GLM-5.2 is zai-org's flagship large language model, specifically engineered for long-horizon tasks with a solid 1M-token context window. This represents a substantial advancement over its predecessor, GLM-5.1, in handling extended inputs and complex, multi-step operations.

Key Capabilities

• 1M-Token Context: Stably sustains long-horizon work, enabling processing of extensive documents and complex conversational histories.
• Advanced Coding: Offers stronger coding capabilities with multiple thinking effort levels, allowing users to balance performance and latency for diverse programming challenges.
• Improved Architecture: Incorporates novel architectural enhancements like IndexShare, which reuses indexers across sparse attention layers to reduce per-token FLOPs by 2.9x at 1M context length. It also features an improved MTP layer for speculative decoding, increasing acceptance length by up to 20%.
• Pure Open-Source: Released under an MIT license, ensuring broad accessibility without regional or technical access restrictions.

Performance Highlights

GLM-5.2 demonstrates strong performance across various benchmarks, particularly in reasoning, coding, and agentic tasks. It shows significant improvements over GLM-5.1 and competes favorably with other leading models in categories like HLE (w/ Tools), CritPt, SWE-bench Pro, NL2Repo, and Terminal Bench 2.1. For instance, it achieves 62.1 on SWE-bench Pro and 81.0 on Terminal Bench 2.1 (Terminus-2).

Good For

• Applications requiring extensive context understanding and generation.
• Complex code generation and debugging with adjustable performance settings.
• Agentic workflows and multi-step reasoning tasks.