Bloom — Hardware Requirements & GPU Compatibility
ChatCode
Bloom is a 176.2B-parameter open language model from BigScience. At Q4_K_M it needs about 116.32 GB of VRAM — see which GPUs and Macs can run it below.
Bloom runs on a 128 GB+ Mac or a datacenter GPU
Best local pick → Mac Pro M2 Ultra (128 GB) · Q4_K_M · 122.4 GB · ~5 tok/s
Macs from 96 GB (Q2_K · ~7 tok/s)
bigscience-bloom-rail-1.0
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Bloom
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
vllm serve bigscience/bloom --max-model-len 16384 --tensor-parallel-size 8vLLM repo with its own quantisation
Download: Weights on Hugging Face
Other ways
each opens step-by-stepRun in cloud
Fits on B200 (192 GB) (69 GB headroom) · Q4_K_M
- Generation speed
- ~45 tok/s
- generation speed
- Cost per 1M output tokens
- $37.16
- per 1M output tokens
How Much VRAM Does Bloom Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | Runs on | |
|---|---|---|---|---|
| 82.4 GB | 95.5 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Runs on RTX PRO 6000 Blackwell Max-Q Workstation Edition 96 GB | |
| 94.5 GB | 107.6 GB | 3-bit medium — 3-bit medium quantization | Runs on Instinct MI250X 128 GB | |
| 116.3 GB | 129.4 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Runs on Instinct MI250X 128 GB | |
| 138.1 GB | 151.2 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Runs on B200 192 GB | |
| 159.9 GB | 173 GB | 6-bit, very good — 6-bit quantization, very good quality | Runs on B200 192 GB | |
| 193.9 GB | 206.9 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Runs on Instinct MI325X 256 GB | |
| 387.7 GB | 400.8 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Runs on 6× H100 SXM 80 GB |
Selected Q4_K_M
4-bit medium quantization122.4GBat 16K context
- Weights
- 115.2 GB
- KV cache
- 7.0 GB
- Overhead
- 0.3 GB
Server-class: ~122.4 GB at 16K context · one Instinct MI250X 128 GB
Which Hardware Can Run Bloom?
Q4_K_M · 116.3 GBBloom (Q4_K_M) requires 116.3 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 152+ GB is recommended. No consumer GPU has enough memory. 17 devices with unified memory can run Bloom, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Pro M2 Ultra (128 GB).
96 GB
Q4_K_M · best fit Q2_K (88.5 GB)
3 GPUs · 2 Mac classes
RTX PRO 6000 Blackwell Max-Q Workstation Edition~14Simulate output on RTX PRO 6000 Blackwell Max-Q Workstation EditionRTX PRO 6000 Blackwell Workstation Edition~14Simulate output on RTX PRO 6000 Blackwell Workstation EditionRTX PRO 6000 Blackwell Server Edition~13Simulate output on RTX PRO 6000 Blackwell Server EditionM3 Ultra*96 GB~7Simulate output on Mac Studio (M3 Ultra, 96GB)M3 Max*96 GB~3Simulate output on MacBook Pro 14" M3 Max (96 GB)128–144 GB
fits · 5.6 GB free · up to 16K ctx
4 GPUs · 2 Mac classes · 1 device family
GH200 Grace Hopper Superchip~27Simulate output on GH200 Grace Hopper SuperchipH200 NVL~27Simulate output on H200 NVLH200 SXM~27Simulate output on H200 SXMM2 Ultra*128 GB~5Simulate output on Mac Pro M2 Ultra (128 GB)M5 Max128 GB~4Simulate output on Mac Studio M5 Max (128 GB)GB10~2Simulate output on ASUS Ascent GX10192+ GB
fits · 69.6 GB free · up to 32K ctx
6 GPUs · 1 Mac class · 2 device families
B200~45Simulate output on B200B300~45Simulate output on B300Instinct MI350X~41Simulate output on Instinct MI350XM3 Ultra*256–512 GB~5Simulate output on Mac Studio (M3 Ultra, 256GB)8x H100 SXM5~150Simulate output on NVIDIA DGX H1008x A100 SXM4~91Simulate output on NVIDIA DGX A100 640GB
By VRAM tier · typical GPUs and Macs at Q4_K_M
Consumer GPUs (up to 32 GB) and Macs up to 64 GB: too small for this model (needs ≥ 88.5 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →
How good is Bloom?
llmrun Score blends public benchmark boards into one number; 100 = the average open model.
Not enough data yet for an llmrun Score.
A Score needs results on enough independent boards. How the Composite Score is built
Benchmarks
Here are all 3 boards that have results for this model, sorted by their weight in the llmrun Score; badges show who ran the test.
Similar models
2| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Qwen3.8 Flash NextAlibaba | — | 109 GB | —needs > 24 GB | same size |
| DeepSeek V4 Flash DSparkDeepSeek | — | 99.6 GB | —needs > 24 GB | same size |
Specifications
- Publisher
- BigScience
- Parameters
- 176.2B
- Architecture
- BloomForCausalLM
- Vocabulary Size
- 250,880
- Release Date
- 2022-05-19
- License
- bigscience-bloom-rail-1.0
Frequently Asked Questions
How much VRAM does Bloom need?
Bloom requires 116.3 GB of VRAM at Q4_K_M, or 387.7 GB at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 176.2B × 4.8 bits ÷ 8 = 105.7 GB
KV Cache + Overhead ≈ 10.6 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
VRAM usage by quantization
Can NVIDIA GeForce RTX 5090 run Bloom?
No — Bloom requires at least 82.4 GB at Q2_K, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
What's the best quantization for Bloom?
For Bloom, Q4_K_M (116.3 GB) offers the best balance of quality and VRAM usage. Q5_K_M (138.1 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 82.4 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Bloom on a Mac?
Yes — Mac Studio (M3 Ultra, 96GB) and 20 other Macs can run Bloom, the smallest at Q2_K. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 116.3 GB of usable unified memory (RAM minus macOS overhead), from Mac Pro M2 Ultra (128 GB).
Can I run Bloom locally?
Yes — Bloom can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Bloom?
At Q4_K_M, Bloom can reach ~41 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 116.3 × 0.65 = ~45 tok/s
Estimated speed at Q4_K_M (116.3 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Bloom?
At Q4_K_M, the download is about 105.75 GB. The full-precision BF16 version is 352.49 GB. The smallest option (Q2_K) is 74.91 GB.
Which GPUs can run Bloom?
No single consumer GPU has enough VRAM to run Bloom at Q4_K_M (116.3 GB). Multi-GPU or professional hardware is required.
Which devices can run Bloom?
17 devices with unified memory can run Bloom at Q4_K_M (116.3 GB), including ASUS Ascent GX10, Mac Pro M2 Ultra (128 GB), Mac Pro M2 Ultra (192 GB), Mac Studio (M3 Ultra, 256GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.