GLM 5v
#149 in Frontier-Spraakmodellez-ai · v5v · siet 1. April 2026 · 11× · tolest 30. Juni 2026
6
Momentum
GLM-5V-Turbo is Z.ai's (Zhipu AI) first native multimodal coding foundation model, built for vision-based coding and agent-driven tasks. It natively processes images, video, document layouts and text, converting design mockups directly into runnable frontend code. The model is deeply optimized for agentic workflows (e.g., with Claude Code and OpenClaw) and, per vendor-reported figures, scores 94.8 on the Design2Code benchmark, while being offered only via API (proprietary, not open-weight). It is built on a 744B-parameter MoE architecture (40B active per token) with a roughly 200K token context window.
Momentum-Verloop
04.04.03.07.
Features
| Key Benchmark (%) | 94.8 on Design2Code (vs. Claude Opus 4.6: 77.3; vendor-reported) |
| Context Window (Tokens) | 202,752 tokens (max. output 131,072 tokens) |
| License | Proprietary/closed (API access only, no open weights, unlike the MIT-licensed GLM-5 base model) |
| Multimodality | Natively multimodal: text, image, video, documents/files as input; text as output |
| Platform | API (OpenAI-compatible) via z.ai, chat.z.ai, OpenRouter |
| Price per 1M Tokens | $1.20 input / $4.00 output per 1M tokens |
| Release Date | April 1, 2026 |