Can 2x RTX Pro 6000 Max-Q (Blackwell, 96 GB, 300 W) run GLM-5.3 Flash?
Yes. Measured 177.7 tok/s generation (vLLM (karmic-kraken image), TP 2, MTP 3 speculative tokens, modelopt_mixed (ModelOpt mixed precision), FP8 KV; source).
Every measured result for GLM-5.3 Flash on 2x RTX Pro 6000 Max-Q (Blackwell, 96 GB, 300 W)
| Build | Quant / setup | Engine | tok/s | |
|---|---|---|---|---|
| 2x RTX PRO 6000 Max-Q (192 GB): GLM-5.3 Flash at 177.7 tok/s single stream (vLLM, MTP) | modelopt_mixed (ModelOpt mixed precision), FP8 KV | vLLM (karmic-kraken image), TP 2, MTP 3 speculative tokens | 177.7 | source |
Memory: 2x RTX Pro 6000 Max-Q (Blackwell, 96 GB, 300 W) has 192 GB. GLM-5.3 Flash weights are about 195.8 GB at Q4_K_M (320B params × 4.89 bits ÷ 8, before context/KV cache). Arithmetic, not a measurement; long contexts need more. Check other quants in the VRAM calculator. At Q4_K_M the weights are about 195.8 GB, more than 192 GB, so Q4_K_M doesn't fully fit without offloading part of the model to system RAM.
Speeds are decode (generation) tokens/sec as reported by each source; quant, engine and context differ, so small gaps are noise.
All measured speeds for GLM-5.3 Flash: best home rig for GLM-5.3 Flash.
GLM-5.3 Flash on other hardware: 2x DGX Spark (GB10) · 4x DGX Spark (GB10) · RTX 5090
Not enough VRAM? Rent before you buy. Run GLM-5.3 Flash on a bigger GPU on a rented 24–96 GB cloud GPU by the hour and see the real speed before spending €1,000+. Try it on Runpod
Referral link: if you sign up with Google and add $10, Runpod gives you and RigBase bonus credit (terms). It never affects our data.
Referral link: if you sign up with Google and add $10, Runpod gives you and RigBase bonus credit (terms). It never affects our data.