GLM-5
Zhipu AI (Z.ai) · 2026-02-12 · 753.9B parameters
GLM-5 is Z.ai's February 12, 2026 open-weight MoE language model for complex systems engineering and long-horizon agentic tasks. Its released checkpoint uses 78 transformer layers, MLA with DeepSeek Sparse Attention, a 202,752-token context window, and turn-level thinking controls; Z.ai describes it as 744B total/40B active and reports 28.5T tokens across base-model pre-training and mid-training. The weights are released under the MIT license.
Benchmark scores
| Benchmark | Score |
|---|---|
| A-Fantasia | 66.7 |
| A-Fantasia Backwards Spelling | 78.0 |
| A-Fantasia Chess | 52.0 |
| A-Fantasia Cube Rotation | 70.0 |
| AA-LCR | 63.3 |
| AA-Omniscience Index | 2.0 |
| AHK-Eval | 80.6 |
| AHK-Eval Algorithms | 83.3 |
| AHK-Eval Data Structures | 66.7 |
| AHK-Eval Date and Time | 83.3 |
| AHK-Eval Easy | 83.3 |
| AHK-Eval Hard | 66.7 |
| AHK-Eval Hidden Cases | 71.8 |
| AHK-Eval Mid | 58.3 |
| AHK-Eval Numbers | 83.3 |
| AHK-Eval Parse Success | 97.2 |
| AHK-Eval Regex | 50.0 |
| AHK-Eval Strings | 50.0 |
| AIME 2024 (avg@8) | 93.3 |
| AIME 2024-2025 | 91.7 |
| AIME 2025 | 96.7 |
| AIME 2025 (avg@8) | 90.0 |
| AIME 2026 | 96.7 |
| ALE-Bench | 1194 |
| Apex (MathArena) | 10.9 |
| APEX (Mercor) | 7.3 |
| Apex Shortlist | 68.6 |
| APEX-Agents | 17.2 |
| APEX-Agents-AA | 14.5 |
| APEX-v1-extended | 49.0 |