FitMyLLM
AI21 Labs/Mixture of Experts

AI21 LabsJamba 1.5 Mini 52B

AI21's Jamba 1.5 Mini — hybrid SSM-Transformer MoE. Fast with 256K context window.

chatreasoningThinkingTool Use
51.6B
Parameters (12B active)
256K
Context length
14
Benchmarks
14
Quantizations
20K
HF downloads
Architecture
MoE
Released
2024-08-22
Layers
32
KV Heads
8
Head Dim
128
Family
jamba

Quantization Options

Context length:
QuantBitsVRAM @ 16KQuality
IQ3_XXS3.25
21.6 GB
21.5 + 0.2 KV
low
IQ3_XS3.5
23.3 GB
23.1 + 0.2 KV
low
Q3_K_S3.64
24.2 GB
24.0 + 0.2 KV
low
IQ3_M3.76
24.9 GB
24.7 + 0.2 KV
low
Q3_K_M4
26.5 GB
26.3 + 0.2 KV
low
Q3_K_L4.3
28.4 GB
28.2 + 0.2 KV
moderate
IQ4_XS4.46
29.4 GB
29.3 + 0.2 KV
moderate
Q4_K_S4.67
30.8 GB
30.6 + 0.2 KV
moderate
Q4_K_M4.89
32.2 GB
32.0 + 0.2 KV
good
Q5_K_S5.57
36.6 GB
36.4 + 0.2 KV
good
Q5_K_M5.7
37.4 GB
37.3 + 0.2 KV
good
Q6_K6.56
43.0 GB
42.8 + 0.2 KV
excellent
Q8_08.5
55.5 GB
55.3 + 0.2 KV
lossless
FP1616
103.9 GB
103.7 + 0.2 KV
lossless

Select your GPU above to see speed estimates and compatibility for each quantization.

Deploying for a team or in production? Size GPUs, cost & scaling in Enterprise →
Ready to run this?Rent by the hour

Rent a GPU and run Jamba 1.5 Mini 52B now

Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.

Community Ratings

Loading ratings...

Benchmarks (14)

Run this model

Easiest way to get started·Beginners
Docs ↗
curl -fsSL https://ollama.com/install.sh | sh
$ollama run hf.co/bartowski/AI21-Jamba-1.5-Mini-GGUF:Q4_K_M

Not in Ollama's library: Ollama pulls the GGUF from Hugging Face (bartowski/AI21-Jamba-1.5-Mini-GGUF). If Q4_K_M is not in that repo, pick a file listed there.

Setup guide
>_

Auto-setup with fitmyllm CLI

Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.

pip install fitmyllmthen run fitmyllmLearn more
Auto-detect GPULive tok/s in chatSpeed benchmarks9 inference engines

GPUs that can run this model

At Q4_K_M quantization. Sorted by minimum VRAM.

Apple M3 Max (36GB)
36 GB VRAM • 300 GB/s
APPLE
$2499
Apple M3 Pro (36GB)
36 GB VRAM • 150 GB/s
APPLE
$1999
NVIDIA A100 PCIe 40GB
40 GB VRAM • 1555 GB/s
NVIDIA
$10000
NVIDIA A100 PCIe 40 GB
40 GB VRAM • 1560 GB/s
NVIDIA
$10000
NVIDIA A100 SXM4 40 GB
40 GB VRAM • 1560 GB/s
NVIDIA
$10000
NVIDIA A800 PCIe 40 GB
40 GB VRAM • 1560 GB/s
NVIDIA
Apple M3 Max (48GB)
48 GB VRAM • 400 GB/s
APPLE
$2899
Apple M4 Pro (48GB)
48 GB VRAM • 273 GB/s
APPLE
$1799
Apple M4 Max (48GB)
48 GB VRAM • 546 GB/s
APPLE
$2499
NVIDIA L40S 48GB
48 GB VRAM • 864 GB/s
NVIDIA
$7500
NVIDIA L40 48GB
48 GB VRAM • 864 GB/s
NVIDIA
$5500
NVIDIA RTX 6000 Ada 48GB
48 GB VRAM • 960 GB/s
NVIDIA
$6800
NVIDIA A40 48GB
48 GB VRAM • 696 GB/s
NVIDIA
$4650
NVIDIA RTX A6000 48GB
48 GB VRAM • 768 GB/s
NVIDIA
$4650
NVIDIA Quadro RTX 8000
48 GB VRAM • 672 GB/s
NVIDIA
NVIDIA Quadro RTX 8000 Passive
48 GB VRAM • 624 GB/s
NVIDIA
NVIDIA A40 PCIe
48 GB VRAM • 696 GB/s
NVIDIA
NVIDIA RTX 6000 Ada Generation
48 GB VRAM • 960 GB/s
NVIDIA
$6800
NVIDIA L20
48 GB VRAM • 864 GB/s
NVIDIA
AMD Radeon PRO W7800 48 GB
48 GB VRAM • 864 GB/s
AMD
$3499
AMD Radeon PRO W7900
48 GB VRAM • 864 GB/s
AMD
$3999
Intel Data Center GPU Max 1100
48 GB VRAM • 1230 GB/s
INTEL
NVIDIA RTX 5880 Ada Generation
48 GB VRAM • 864 GB/s
NVIDIA
$5500
NVIDIA RTX PRO 5000 Blackwell
48 GB VRAM • 1340 GB/s
NVIDIA
$4999
AMD Radeon PRO W7900D
48 GB VRAM • 864 GB/s
AMD
$3999
NVIDIA GRID A100B
48 GB VRAM • 1870 GB/s
NVIDIA
NVIDIA RTX A6000
48 GB VRAM • 768 GB/s
NVIDIA
$4650
NVIDIA L40
48 GB VRAM • 864 GB/s
NVIDIA
$7000
NVIDIA L40S
48 GB VRAM • 864 GB/s
NVIDIA
$8000
Apple M5 Pro (48GB)
48 GB VRAM • 200 GB/s
APPLE

Find the best GPU for Jamba 1.5 Mini 52B

Build Hardware for Jamba 1.5 Mini 52B

AI21's Jamba 1.5 Mini — hybrid SSM-Transformer MoE. Fast with 256K context window.

Spec sheet

Jamba 1.5 Mini 52B — 51.6B MoE.

Specifications
Parameters
51.6B (12B active)
Architecture
Mixture of Experts
Context Length
256K tokens
Capabilities
chat, reasoning
Release Date
2024-08-22
Provider
AI21 Labs
Family
jamba
VRAM requirements
QuantBPWVRAMQuality
IQ3_XXS3.2521.5 GB82%
IQ3_XS3.523.1 GB84%
Q3_K_S3.6424.0 GB85%
IQ3_M3.7624.7 GB86%
Q3_K_M426.3 GB88%
Q3_K_L4.328.2 GB90%
IQ4_XS4.4629.3 GB92%
Q4_K_S4.6730.6 GB93%
Q4_K_M4.8932.0 GB94%
Q5_K_S5.5736.4 GB96%
Q5_K_M5.737.3 GB96%
Q6_K6.5642.8 GB97%
Q8_08.555.3 GB100%
FP1616103.7 GB100%
Benchmark scores
HumanEval64.0
MMLU-PRO42.0
MATH57.0
IFEval64.0
BBH10.7
GPQA2.5
MUSR3.7
GPQA Diamond30.2
LiveCodeBench6.2
AIME1.0
MATH-50035.7
HLE5.1
AA Intelligence8.0
SciCode8.0
Compatible GPUs
30 @ Q4_K_M