GLM-5.3 on local hardware
Estimated comfort for repository coding with Aider on a MacBook Pro M4 Pro 48GB.
GLM-5.3
zai-org · other · 1,834,350 downloads
Mixture of expertsReasoning / thinking mode
- Parameters
- 753B exact
- Active per token
- ≈50B (8 of 256 experts)
- Context window
- 1M tokens
- Attention
- Multi-head latent attention (compressed KV)
- Layers · KV heads · head dim
- 78 · 64 · 192
- KV cache
- 88 KB/token · 2.7 GB at 32K
- Tool calling
- good (from chat template)
- Thinking
- good (from chat template)
- Vision
- basic (inferred — no supporting signal)
- Commercial use
- check cardCheck the model card
- Import confidence
- medium · capabilities estimated
Weights by quantization
FP8
709 GB
computed · low confidence
- • Uses a compressed / sparse attention scheme; KV-cache size is approximate.
- • Assumes GGUF / MLX conversions of this model exist (they do for most popular models). Sizes are computed from bits per weight.
- • vision: basic, inferred from missing signals (low confidence).
- • Capability ratings are estimated from model size and name — this model hasn't been assessed by hand.
- • Commercial use unknown: Check the model card.
GLM-5.3 — FP8
Does not fit in memory
Needs ≈713 GB but only ≈37 GB is available after the OS, other apps and Aider.
High confidence