Can I run GLM-5.2 on an RTX 4090 (24GB)?
Computed with the open FitLLM engine — accurate per-layer KV-cache modeling, not a naive estimate. Updated 2026-07-10.
Memory breakdown (Q4_K_M, F16 KV, 33K context)
| Model weights | 429 GB |
| KV cache | 2.7 GB |
| Runtime overhead + reserve | 54.9 GB |
| Total used | 487 / 24 GB |
| Short by | 463 GB |
Max context that fits at Q4_K_M: does not fit.
Every quantization on the RTX 4090
| Weight quant | Weights | Fits (KV F16) | Used @32K |
|---|---|---|---|
| Q4_K_M | 429 GB | ❌ won't fit | 487 / 24.0 GB |
| Q5_K_M | 500 GB | ❌ won't fit | 566 / 24.0 GB |
| Q6_K | 575 GB | ❌ won't fit | 651 / 24.0 GB |
| Q8_0 | 745 GB | ❌ won't fit | 841 / 24.0 GB |
| FP16 | 1403 GB | ❌ won't fit | 1577 / 24.0 GB |
Lower weight quants free memory at some output-quality cost — Q4 is the common sweet spot; below that quality drops faster.
KV cache is F16 here (llama.cpp default). Drop it to Q8/Q4 (-ctk/-ctv) for more context.
Embed this verdict
Live badge for your README or model card — recomputed by the engine, never stale:
[](https://fitllm.run/can-i-run/glm-5-2-on-rtx-4090)
Or from your terminal (exit 0/1 — works as a pre-download guard):
npx fitllm "GLM-5.2" --gpu "RTX 4090"Why most VRAM calculators get this wrong
GLM-5.2 uses MLA (Multi-head Latent Attention): K/V are compressed into a single low-rank latent (512 + 64 RoPE dims) shared across all heads — cached once, not per-head K and V. Naive "2 × heads × head_dim × layers" formulas over-count its KV cache by an order of magnitude.
What fits on the RTX 4090 instead
same GPU Models that fit on the RTX 4090: GLM-4.7-Flash, gpt-oss-20b, Qwen 3.6 27B, Gemma 4 e2b, Gemma 4 e4b, Gemma 4 12b, Gemma 4 26b A4B, Llama-3.2-3B-Instruct, Llama-3.1-8B-Instruct, Qwen3-0.6B, Qwen3-1.7B, Llama-3.2-1B-Instruct, Gemma-3-1B-it.
same model GPUs that run GLM-5.2: —.
Reproduce it
GLM-5.2 = 753B (40B active, MoE), 78 layers. The RTX 4090 has 24GB / 1008GB/s. Same math, open source: fitllm-engine. GGUF bpw from llama.cpp.
All numbers are computed by the open-source fitllm-engine (MIT) from official model config.json values — reproduce or audit them yourself. Estimates; real usage varies with runtime (llama.cpp / MLX / Ollama), driver and display. Found a mismatch? Report it. · FitLLM home