FitMyLLM
AI21/Mixture of Experts

AJamba Large 1.6

The AI21 Jamba 1.6 family of models is state-of-the-art, hybrid SSM-Transformer instruction following foundation models.

chatreasoningtool_usemultilingual
398B
Parameters (94B active)
256K
Context length
1
Benchmarks
17
Quantizations
0
Architecture
MoE
Released
2025-09-20
Layers
72
KV Heads
8
Head Dim
128
Family
jamba

Quantization Options

Context length:
QuantBitsVRAM @ 16KQuality
IQ2_XXS2.38
122.3 GB
118.9 + 3.4 KV
low
IQ2_M2.93
149.6 GB
146.3 + 3.4 KV
low
Q2_K3.16
161.1 GB
157.7 + 3.4 KV
low
IQ3_XXS3.25
165.6 GB
162.2 + 3.4 KV
low
IQ3_XS3.5
178.0 GB
174.6 + 3.4 KV
low
Q3_K_S3.64
185.0 GB
181.6 + 3.4 KV
low
IQ3_M3.76
190.9 GB
187.5 + 3.4 KV
low
Q3_K_M4
202.9 GB
199.5 + 3.4 KV
low
Q3_K_L4.3
217.8 GB
214.4 + 3.4 KV
moderate
IQ4_XS4.46
225.7 GB
222.4 + 3.4 KV
moderate
Q4_K_S4.67
236.2 GB
232.8 + 3.4 KV
moderate
Q4_K_M4.89
247.1 GB
243.8 + 3.4 KV
good
Q5_K_S5.57
281.0 GB
277.6 + 3.4 KV
good
Q5_K_M5.7
287.4 GB
284.1 + 3.4 KV
good
Q6_K6.56
330.2 GB
326.8 + 3.4 KV
excellent
Q8_08.5
426.7 GB
423.4 + 3.4 KV
lossless
FP1616
799.9 GB
796.5 + 3.4 KV
lossless

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

Too big for a single GPU — plan a multi-GPU deployment
Even the lightest quant needs ~122 GB. Size GPUs, replicas, TCO and scaling for a production setup. Open in Enterprise →
READY TO RUN THIS?RENT BY THE HOUR

RENT A GPU AND RUN JAMBA LARGE 1.6 NOW

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

Community Ratings

Loading ratings...

Benchmarks (1)

AA Intelligence11.0

Run this model

Easiest way to get started·Beginners
DOCS ↗
curl -fsSL https://ollama.com/install.sh | sh
$ollama run jamba:398b-q4_K_M

Tag may need adjustment — check ollama.com/library/jamba for available tags.

▸ 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.

AMD Radeon Instinct MI325X
288 GB VRAM • 10300 GB/s
AMD
$20000
AMD Radeon Instinct MI350X
288 GB VRAM • 8190 GB/s
AMD
$25000
AMD Radeon Instinct MI355X
288 GB VRAM • 8190 GB/s
AMD
$30000
Apple M4 Ultra (384GB)
384 GB VRAM • 1092 GB/s
APPLE
$9999
Apple M5 Ultra (384GB)
384 GB VRAM • 1228 GB/s
APPLE

Find the best GPU for Jamba Large 1.6

Build Hardware for Jamba Large 1.6
▸ SPEC SHEET

Jamba Large 1.6398B MoE.

▸ SPECIFICATIONS
PARAMETERS
398B (94B active)
ARCHITECTURE
Mixture of Experts
CONTEXT LENGTH
256K tokens
CAPABILITIES
chat, reasoning, tool_use, multilingual
RELEASE DATE
2025-09-20
PROVIDER
AI21
FAMILY
jamba
▸ VRAM REQUIREMENTS
QUANTBPWVRAMQUALITY
IQ2_XXS2.38118.9 GB65%
IQ2_M2.93146.3 GB75%
Q2_K3.16157.7 GB78%
IQ3_XXS3.25162.2 GB82%
IQ3_XS3.5174.6 GB84%
Q3_K_S3.64181.6 GB85%
IQ3_M3.76187.5 GB86%
Q3_K_M4199.5 GB88%
Q3_K_L4.3214.4 GB90%
IQ4_XS4.46222.4 GB92%
Q4_K_S4.67232.8 GB93%
Q4_K_M4.89243.8 GB94%
Q5_K_S5.57277.6 GB96%
Q5_K_M5.7284.1 GB96%
Q6_K6.56326.8 GB97%
Q8_08.5423.4 GB100%
FP1616796.5 GB100%
§ 01BENCHMARK SCORES
AA Intelligence11.0
§ 03COMPATIBLE GPUs
5 @ Q4_K_M