Skip to content
Bitpute

How much VRAM does DBRX 132B (36B active) Instruct need?

By Bitpute · Published 12 July 2026 · Updated 24 July 2026 · How we estimate · Sources · Editorial policy · Version history · Report an error · Figures re-checked against the calculation engine on every build

DBRX 132B (36B active) Instruct needs about 79.0 GB of VRAM at Q4_K_M, or 137.9 GB at Q8_0 and 258.9 GB at FP16. No consumer card holds it at Q4_K_M. The smallest fit in our set is the H200 141GB (141 GB). Weights plus runtime overhead; KV cache for your context length is on top.

Weights-plus-runtime footprint across the six most common quantizations. KV cache is context-dependent and comes on top. At 32K it is gigabytes on its own, before the weights; size it precisely in the GPU Memory Calculator.

This is a mixture-of-experts model: all 132B parameters must sit in memory, but only about 36B are active per token — so it loads like a 132B model and runs closer to a 36B one.

This is the instruction-tuned checkpoint: the same architecture and the same memory footprint as the base model, fine-tuned to follow prompts and hold a conversation. If you intend to fine-tune on your own data, start from DBRX 132B (36B active) instead.

Parameters132B
Active / token36B
Context32K
VendorDatabricks
LicenseOpen weights (see card)
Released2024

Can your GPU run DBRX 132B (36B active) Instruct?

Pick your card — the answer below is computed by the same engine that produces every figure on this page.

Yes — recommended
Estimated VRAM79.0 GB
GPU VRAM141 GB
Headroom54.9 GB

H200 141GB runs DBRX 132B (36B active) Instruct at Q4_K_M with 54.9 GB to spare; comfortable at this size, where most cards cannot hold the weights at all; as a mixture-of-experts model all 132B parameters must be resident even though only 36B are active per token.

Weights plus runtime overhead. KV cache is not included — it depends on context length and this model’s published architecture.

Why
Weights (Q4_K_M)74.5 GB
Runtime overhead4.5 GB
Estimated total79.0 GB
Usable VRAM (95% of 141 GB)133.9 GB

Across every GPU we track

Estimates, not guarantees: real usage moves with runtime, driver, batch size and context. Size a specific context window in the GPU Memory Calculator.

How was this number calculated? Every figure here comes from Bitpute’s documented calculation methodology — parameters, precision, quantization, runtime overhead and usable VRAM, each formula written out in full.

Memory by quantization

QuantWeights+ runtime overhead
FP16245.9 GB258.9 GB
Q8_0130.6 GB137.9 GB
Q6_K100.8 GB106.6 GB
Q5_K_M87.4 GB92.6 GB
Q4_K_M74.5 GB79.0 GB
Q4_069.2 GB73.4 GB

Overhead = 0.75 GB + 5% of weights (CUDA context, buffers). Add KV cache on top. Weights dominate at this size, but a 32K window still costs gigabytes.

Which quantization should you actually run?

For DBRX 132B (36B active) Instruct, the single-GPU entry point is Q4_K_M (79.0 GB) on H200 141GB, with plenty of headroom for long context. Higher quality (Q6_K 106.6 GB, FP16 258.9 GB) needs a 48 GB+ card or multiple GPUs — for occasional use, renting is usually cheaper than buying. As a mixture-of-experts model it must hold all 132B in VRAM but activates only about 36B per token, so it runs much faster than its footprint suggests — memory is the limit here, not speed.

Single-GPU fit at Q4_K_M (79.0 GB + KV)

RTX 3060 12GBRTX 4060 Ti 16GBRTX 4070 Ti SUPERRTX 3090RTX 4090RTX 5090A100 40GBRTX A6000L40SA100 80GBH100 80GBH200 141GB

Size it exactly →Rent a GPU for it →

Where the memory goes

Memory budget for DBRX 132B (36B active) Instruct on a H200 141GBStacked bar. Weights 74.5 GB, runtime overhead 4.5 GB, KV cache at 8K context 1.0 GB, against 133.9 GB usable on a H200 141GB.Memory budget for DBRX 132B (36B active) Instruct on a H200 141GB133.9 GB usable of 141 GBWeights (Q4_K_M) — 74.5 GBRuntime overhead — 4.5 GBKV cache @ 8K — 1.0 GBHeadroom — 53.9 GB

Weights and overhead are exact for Q4_K_M. The KV bar is a generic 32-layer transformer at 8K; at this scale the weights dominate, but long context still adds gigabytes — size it in the calculator is for.

Common questions

How much VRAM does DBRX 132B (36B active) Instruct need?

DBRX 132B (36B active) Instruct needs about 79.0 GB of VRAM at Q4_K_M, 137.9 GB at Q8_0, or 258.9 GB at FP16. That is weights plus about 0.75 GB of runtime overhead and 5% of weight size; KV cache is additional and depends on context length.

What GPU can run DBRX 132B (36B active) Instruct?

At Q4_K_M the smallest card in our set that fits is the H200 141GB (141 GB usable at 95%). At FP16 no single GPU in our set is sufficient.

Can DBRX 132B (36B active) Instruct run on 24 GB?

Not at Q4_K_M: it needs about 79.0 GB, more than the ~22.8 GB usable on a 24 GB card. Use a larger card, multiple GPUs, or offload.

Same memory footprint as 1 other checkpoint

DBRX 132B (36B active) Instruct has an identical parameter count to this, so every figure on this page applies to it unchanged. What varies is different checkpoint roles (base) — not the size. Choose on capability and licence; the memory budget is identical.

DBRX 132B (36B active)

More in the DBRX family

DBRX 132B (36B active)