RTX 3090 x8
4recorded runs
91.2best tok/s
4sourced
How it compares — 24 GB per card · multi-GPU
What fits on 24 GB — measured
| Model | Quant | Peak VRAM | Headroom |
|---|---|---|---|
| DeepSeek-V4-Flash-0731 | FP8_E4M3 | 184.3 GB | -160.3 GB |
| GLM-5.3-Flash | AWQ-4bit | 183.4 GB | -159.4 GB |
| keys-DeepSeekV4Flash-Vision-EXP-ablit | FP8 | 183.1 GB | -159.1 GB |
| GLM-5.3-Flash | Unsloth-Dynamic-IQ4_XS | 165.9 GB | -141.9 GB |
All benchmarks Compare RTX 3090 x8 →
| GPU | Model | Quant | Backend | Ctx | tok/s | prompt tok/s | VRAM GB | Source |
|---|---|---|---|---|---|---|---|---|
| RTX 3090 x8 · 24 GB | DeepSeek-V4-Flash-0731 | FP8_E4M3 | vllm 0.11.2.dev278+gccfeb3a8 (LimeChain s… | 327680 | 91.2 ✓ | 1344.2 | 184.3 | src |
| RTX 3090 x8 · 24 GB | GLM-5.3-Flash | AWQ-4bit | vllm 0.28.1rc1.dev125+g85872952a.glm53sm8… | 204800 | 75.2 ✓ | 1020.0 | 183.4 | src |
| RTX 3090 x8 · 24 GB | keys-DeepSeekV4Flash-Vision-EXP-ablit | FP8 | vllm 0.11.2.dev278+gccfeb3a8 (LimeChain s… | 131072 | 66.8 ✓ | 1088.3 | 183.1 | src |
| RTX 3090 x8 · 24 GB | GLM-5.3-Flash | Unsloth-Dynamic-IQ4_XS | llama.cpp b10667-2e0e57f1 | 131072 | 42.1 ✓ | 148.4 | 165.9 | src |