GLM-4-32B-0414
- Open weights
Prices
1 offering · prices as listedTHUDM/GLM-4-32B-0414| Conditions/ 1M tokens | Input | Output |
|---|---|---|
| Standard | ¥1.89 | ¥1.89 |
Prices from 硅基流动：模型价格 (opens in a new tab) (checked Nov 10, 2026)
Standard: the price a request with no conditions pays (the headline price elsewhere on the site).
Details
- Family
- glm
- Model ID
Run locally
Weights: zai-org/GLM-4-32B-0414- Parameters 32.6B
- Dense
- Native precision BF16
- Layers 61
- KV cache per token 61 KB
- Config positions 32K (config.json max_position_embeddings)
- Needs
- ≈ 18.4 GB (weights 17.1 + KV cache 0.5 + overhead 0.9)
- Usable
- 21.6 GB (90% of its memory)
- Generation speed
- 27–40 tok/s
- Longest context
- 32K (the model’s limit)
- First token
- about 1.05–1.58 s for a 2K prompt
| Precision | Weights | Needs at 8K | On your hardware | Longest context |
|---|---|---|---|---|
| BF16 (native) | 60.7 GB | ≈ 64.2 | ✕Doesn’t fit | — |
| Q8_0 | 32.2 GB | ≈ 34.3 | ○Needs system memory 3–4 tok/s | — |
| FP8 | 30.3 GB | ≈ 32.3 | ○Needs system memory 3–5 tok/s | — |
| Q6_K | 24.9 GB | ≈ 26.6 | ○Needs system memory 6–9 tok/s | — |
| Q5_K_M | 21.6 GB | ≈ 23.1 | ○Needs system memory 12–18 tok/s | — |
| Q5_0 | 20.9 GB | ≈ 22.4 | ○Needs system memory 15–22 tok/s | — |
| Q4_K_M | 18.4 GB | ≈ 19.8 | ●Runs well 25–37 tok/s | 32K |
| Q4_0 | 17.1 GB | ≈ 18.4 | ●Runs well 27–40 tok/s | 32K |
| IQ4_XS | 16.1 GB | ≈ 17.4 | ●Runs well 28–42 tok/s | 32K |
| Q3_K_Mquality drops | 14.8 GB | ≈ 16.0 | ●Runs well 31–46 tok/s | 32K |
| IQ3_XXSquality drops | 11.6 GB | ≈ 12.7 | ●Runs well 39–58 tok/s | 32K |
| Q2_Kquality drops | 11.4 GB | ≈ 12.4 | ●Runs well 40–59 tok/s | 32K |
| IQ2_XXSquality drops | 7.8 GB | ≈ 8.8 | ●Runs well 57–85 tok/s | 32K |
Click a row to choose a precision. “Maker’s file” sizes are the maker’s own files; others are parameters × bits per weight.
Compared with other hardware
Each at the highest precision it runs. Click one for details.
Smallest that runs it (Q4_K_M, 8K): GPU GeForce RTX 4090; unified memory Apple M6 (24GB or 32GB) · 32 GB.
Run commands
The maker publishes no files at this precision. Choose its native precision or one of its own builds, or quantise it yourself with llama.cpp.
Memory to fine-tune
Batch 1, 2,048-token sequences, gradient checkpointing; LoRA and QLoRA train rank-16 adapters on the attention projections.
Estimates; facts from the maker’s Hugging Face repositories, as of 2026-11-11. How it’s estimated
Cited scores
From third-party leaderboards, not measured by us- EQ-Bench Creative Writing v3 1,067
#81 of 94 listed models
- EQ-Bench by Samuel J. Paech (eqbench.com; EQ-bench/EQ-bench-site on GitHub), MIT License. · License · as of Oct 28, 2026
- changes we made
Changes
All changes →- New language model
Sources
- zai-org/GLM-4-32B-0414 model card (opens in a new tab) huggingface.co · checked Nov 10, 2026
- 硅基流动：模型价格in Chinese (opens in a new tab) siliconflow.cn · checked Nov 10, 2026
- zai-org/GLM-4-32B-0414 config.json (opens in a new tab) huggingface.co · checked Nov 10, 2026
Zhipu AI’s official policies: data use, privacy, terms →
Some values on this page are still waiting to be checked against official pages: a filled dot is verified, a hollow one is not yet (with or without a source on file). How we verify data
Zhipu AI · China · z.ai
Related models
Snapshots, the same family and other models from this provider