selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

internlm logo

Can the RTX A4000 run InternLumina-U2?

InternLumina-U2·16B·RTX A4000·16 GB VRAM

Saved in this browser, used on every model page.
Yes

Yes. The RTX A4000 runs InternLumina-U2 fully in VRAM at Q6_K (about 15 GB of 16 GB), at roughly 19 tokens/s.

best quant
Q6_K
needs
15 GB
usable
16 GB
est. speed
~19 tok/s

Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the InternLumina-U2 page →

Every quantisation of InternLumina-U2 on the RTX A4000

QuantQualityMemoryFitSpeed
Q2_K Heavy loss7.1 GBfits~47 tok/s
Q3_K_M Noticeable loss9.6 GBfits~32 tok/s
Q4_K_M Small loss11 GBfits~25 tok/s
Q5_K_M Near-lossless13 GBfits~22 tok/s
Q6_K pickNear-lossless15 GBfits~19 tok/s
Q8_0 Lossless19 GBoffload~7.6 tok/s
FP16 / BF16 Lossless33 GBoffload~1.8 tok/s

Speed is a bandwidth-based estimate for generating text (448 GB/s); prompt processing and long contexts are slower.

Measured speed on the RTX A4000

No community reports for this pair yet.

Report your own speed →

Omni models that fit the RTX A4000 comfortably

All that fit →
ModelParamsBest quantMemorySpeedLicence
12BQ6_K≈13 GBFits in memory~25 tok/sCommercial OK
ByteDance-Seed logoBAGEL-7B-MoTByteDance-Seed
14.7BQ6_K≈13 GBFits in memory~20 tok/sCommercial OK
17.5BQ6_K≈9.3 GBFits in memory~17 tok/sCommercial OK
10.7BQ8_0≈8.9 GBFits in memory~22 tok/sCommercial OK
12BQ6_K≈13 GBFits in memory~25 tok/sCommercial OK
9.4BQ8_0≈10 GBFits in memory~25 tok/sCommercial OK

Other GPUs that run InternLumina-U2

Frequently asked questions

Can the RTX A4000 run InternLumina-U2?
Yes. The RTX A4000 runs InternLumina-U2 fully in VRAM at Q6_K (about 15 GB of 16 GB), at roughly 19 tokens/s.
How fast does InternLumina-U2 run on the RTX A4000?
An estimate from memory bandwidth gives about 19 tokens/s at Q6_K.
Which quantisation of InternLumina-U2 should I use on the RTX A4000?
Q6_K: the largest one that fits. Expected quality: near-lossless compared with the full-precision weights.

Memory and speed are estimates from the model's size and the card's memory bandwidth; real numbers depend on the runtime, drivers and context length. Community figures are medians of user reports.