ibm-granite · text · mixture of experts

granite-3.1-1b-a400m-instruct

ibm-granite/granite-3.1-1b-a400m-instruct

granite-3.1-1b-a400m-instruct at Q4_K_M is exactly 821,847,360 bytes (0.77 GiB / 0.82 GB) — an effective 4.926 bits per weight, not the nominal 4. Its KV cache at 32K is 1.50 GiB.

From the file· summed from 1 file(s)From the file· KV per layer
Parameters
1.3B
total, not active
Architecture
granitemoe
24 layers
Context
131,072
native (config.json)
License
apache-2.0

Shipped quantizations

● exact bytes, summed from published files
QuantSize●Exact bytes●Effective bpw●Tensors●Publisher
Q2_K0.48 GiB512,238,9123.071—bartowski
Q2_K_L0.49 GiB524,429,3443.143—bartowski
IQ3_XS0.53 GiB567,780,6723.403—bartowski
Q3_K_S0.56 GiB597,861,6963.584—bartowski
IQ3_M0.57 GiB609,559,8723.654—bartowski
Q3_K_M0.61 GiB658,580,8003.948—bartowski
Q3_K_L0.66 GiB711,402,8164.264—bartowski
IQ4_XS0.68 GiB728,212,8004.365—bartowski
IQ4_NL0.72 GiB768,320,8324.606—bartowski
Q4_00.72 GiB771,466,5604.624—bartowski
Q4_K_S0.72 GiB774,874,4324.645—bartowski
Q4_K_M0.77 GiB821,847,3604.926—bartowski
Q4_K_L0.78 GiB834,037,7924.999—bartowski
Q5_K_S0.86 GiB928,752,9605.567—bartowski
Q5_K_M0.89 GiB956,327,2325.732—bartowski
Q5_K_L0.90 GiB968,517,6645.806—bartowski
Q6_K1.02 GiB1,099,212,0966.589—bartowski
Q6_K_L1.04 GiB1,111,402,5286.662—bartowski
Q8_01.32 GiB1,422,239,7768.525—bartowski
F162.49 GiB2,672,669,21616.020—bartowski

KV cache by context

computed per layer
ContextKV cache (f16)●Flat formulaOverstated byFull / windowed / recurrent
4,0960.19 GiB0.19 GiB—24 / 0 / 0
8,1920.38 GiB0.38 GiB—24 / 0 / 0
16,3840.75 GiB0.75 GiB—24 / 0 / 0
32,7681.50 GiB1.50 GiB—24 / 0 / 0
65,5363.00 GiB3.00 GiB—24 / 0 / 0
131,0726.00 GiB6.00 GiB—24 / 0 / 0

Compare with

same modality, comparable size
vs Qwen3-4Bvs HyperCLOVAX-SEED-Text-Instruct-1.5Bvs Llama-3.2-1B-Instructvs ced-basevs gemma-3-1b-itvs Qwen3-1.7BAll alternatives →Will it fit on my card? →

Will it run on your card?

full quant x context sweep
By VRAM:8GB12GB16GB24GB32GB48GB
Radeon RX 6500 XT 4GBGeForce RTX 3050 6GBGeForce RTX 5050 8GBGeForce RTX 3080 10GBGeForce RTX 2080 Ti 11GBGeForce RTX 5070 12GBGeForce RTX 5060 Ti 16GBApple M3 Pro 18GBGeForce RTX 3080 Ti 20GBGeForce RTX 5090 D V2 24GBGeForce RTX 5090 D 32GBApple M5 Max 36GBApple M5 Pro 48GBApple M5 Max 64GBApple M3 Ultra 96GBApple M5 Max 128GBApple M2 Ultra 192GBApple M3 Ultra 256GBApple M3 Ultra 512GB

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 0.70 GiB. The real file is 0.77 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

from config.json
Layers
24
Attention heads
16
KV heads
8
Head dim
64
Hidden size
1024
Vocab
49,155
Sliding window
none
SWA period
—
MLA
no
Experts
32
Experts per token
8
use_sliding_window
—

Questions people ask

How much VRAM does granite-3.1-1b-a400m-instruct need?
Q4_K_M is exactly 821,847,360 bytes (0.77 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is granite-3.1-1b-a400m-instruct's KV cache?
1.50 GiB at 32K context with an f16 cache, computed per layer. Quantizing the cache to q8_0 roughly halves it, which is often the difference between a context length fitting and not.
Is granite-3.1-1b-a400m-instruct a mixture-of-experts model?
Yes — 32 experts, 8 routed per token. Every expert must be resident, but only the routed ones are read per token, which is why its memory requirement and its speed behave very differently.
Which quantization of granite-3.1-1b-a400m-instruct should I use?
Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.