AMD Instinct MI250: what runs best?

RTX 4090 RTX 5090 RTX 3090 RTX 4060 Ti (16 GB)RTX 5070 Ti Radeon RX 7900 XTX M4 Max 128 GB M3 Ultra 512 GB DGX Spark (128 GB)All devices

AMD · Multi-GPU module

AMD Instinct MI250

Memory128 GB HBM2eBandwidth3,200 GB/s vendor specLogical GPUs2 × 64 GB board totals are not one GPUPlussystem RAM for offload (set below)

GPUs

Instinct MI250128 GB · 3,200 GB/s1

Custom GPU

System RAM

InstalledGBTypeSpeedMT/sChannels× 64-bit

≈ 89.6 GB/s theoretical ceiling (MT/s × 8 bytes × channels). DDR5’s two 32-bit subchannels count as one channel.

CPU

PlatformThreads for inference

Used for CPU layers/experts and the -t flag. A reference platform fills its maximum channels and speed; set what you actually populated.

Advanced: reserves, bandwidth, workload, models

Available VRAM per GPUGiB Runtime allowance per deviceGiB Spare margin per deviceGiB Unified memory for models% RAM kept for OS and appsGiB Measured RAM bandwidthGB/s GPU linkLanes Prompt length (first token)tokens Custom contexttokens Parallel sequences

Allow disk-streamed experts (documented cases)Include base checkpointsInclude specialistsInclude community & quant-trained variantsInclude models with unverified release dates

Defaults are editorial planning choices, not measurements: 1 GiB runtime + 0.5 GiB margin per device, 75% of unified memory, 8 GiB RAM for the OS, a vendor “24 GB” treated as 24 GiB (enter your driver’s available VRAM to override), 5 t/s minimum speed (research editorial default). Memory ledger

2× Instinct MI25064 GB DDR5-5600 ×2 ch · 56.0 GiB usable · 89.6 GB/s theoretical8K contextF16 KVany backend≥ 5 t/s

Best plans 2× Instinct MI250

Best quality at ≥ 5 t/s

No plan with a quality score reaches 5 t/s on this setup.

Faster tradeoff

No faster plan with a quality score.

Possibly better · speed not established

Qwen3.8-Flash-Next

39.8 AA index single mode

Q4_K_Mllama.cpp ROCmLayer split across GPUs

speed Unknown

No report, calibration or bandwidth figure for this setup. Also: Qwen3.8-27B, NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16.

WeightsKV cacheReserveFreeFits* = a state term unsizedMeasuredlinks to its sourceEstimatedcalibrated or extrapolated range, ~Ceilingupper bound, ≤Unknownnever zero

Memory not fully resolved 18

The weights fit, but the KV cache for this attention type (or the chosen cache layout) is not established, so a fit cannot be guaranteed.

1
Ling-3.0-flash-VL
inclusionAI/Ling-3.0-flash-VL
2026-09-09124B MoE · 5.5B active
AA index24.6single mode
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsKV unknown
2 GPUs
126+ / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
2
Ling-3.0-flash-Fin
inclusionAI/Ling-3.0-flash-Fin
2026 (month unknown)124B MoE · 5.1B active
AA index22.6single mode
Plan & memory
Q5_K_Mllama.cpp ROCmLayer split across GPUsKV unknown
2 GPUs
87.4+ / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
3
gemma-4-26B-A4B-it
google/gemma-4-26B-A4B-it
2026-04-0225B MoE · 3.8B active
AA index16.7 est.Reasoning
Plan & memory
native BF16vLLMFull GPUKV unknown
GPU
49.6+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
4
gemma-4-31B-it
google/gemma-4-31B-it
2026-04-0231B dense
AA index14.7Reasoning
Plan & memory
native BF16vLLMFull GPUKV unknown
GPU
59.8+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
5
gemma-4-12B-it
google/gemma-4-12B-it
2026-06-0312B dense
AA index14.2 est.Reasoning
Plan & memory
native BF16vLLMFull GPUKV unknown
GPU
23.8+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
6
LongCat-Flash-Lite
meituan-longcat/LongCat-Flash-Lite
2026-01 (day unknown)69B MoE · 3B active
AA index11.5 est.single mode
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsKV unknown
2 GPUs
70.9+ / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
7
Mistral-Small-4-119B-2603
mistralai/Mistral-Small-4-119B-2603
2026-03-16119B MoE · 6B active
AA index11.3Reasoning
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsKV unknown
2 GPUs
121+ / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
8
Ling-3.0-tiny
inclusionAI/Ling-3.0-tiny
2026-08 (day unknown)7.9B MoE · 1.3B active
AA index11.1single mode
Plan & memory
FP16llama.cpp ROCmFull GPUKV unknown
GPU
16.2+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
9
Qwen3-Next-80B-A3B-InstructOlder
Qwen/Qwen3-Next-80B-A3B-Instruct
2025-09-1181B tensors · MoE
AA index9.6 est.single mode
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsKV unknown
2 GPUs
82.0+ / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
10
diffusiongemma-26B-A4B-it
google/diffusiongemma-26B-A4B-it
2026-06-1025B MoE · 3.8B active
AA index9.5 est.single mode
Plan & memory
native BF16vLLMFull GPUKV unknown
GPU
49.6+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
11
Qwen3-Coder-NextSpecialist
Qwen/Qwen3-Coder-Next
2026-02-0280B MoE · 3B active
AA index9.2single mode
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsKV unknown
2 GPUs
82.0+ / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
12
gemma-4-E4B-it
google/gemma-4-E4B-it
2026-04-028B dense
AA index8.9Reasoning
Plan & memory
native BF16vLLMFull GPUKV unknown
GPU
16.4+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
13
Falcon-H1R-7B
tiiuae/Falcon-H1R-7B
2026-01-057B dense
AA index7.8 est.single mode
Plan & memory
native BF16vLLMFull GPUKV unknown
GPU
15.6+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
14
gemma-4-E2B-it
google/gemma-4-E2B-it
2026-04-025.1B dense
AA index7.8 est.Reasoning
Plan & memory
native BF16vLLMFull GPUKV unknown
GPU
11.0+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
15
Llama 3.3 70B InstructOlderGated
meta-llama/Llama-3.3-70B-Instruct
2024-12-0671B tensors
AA index7.7 est.single mode
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsKV unknown
2 GPUs
72.8+ / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
16
Llama 4 Scout 17B-16E InstructOlderGated
meta-llama/Llama-4-Scout-17B-16E-Instruct
2025-04-05109B tensors
AA index6.5single mode
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsKV unknown
2 GPUs
109+ / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
17
Phi-4-mini-instructOlder
microsoft/Phi-4-mini-instruct
2025-02-263.8B tensors · dense
AA index6.3 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUKV unknown
GPU
8.7+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
18
tiny-aya-globalGated
CohereLabs/tiny-aya-global
2026-02-173.3B tensors
AA index4.8 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUKV unknown
GPU
7.7+ / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown

+ 30 models without a AA index score

DeepSeek-R1-Distill-Qwen-1.5Bnative BF16 · full gpu
DeepSeek-R1-Distill-Qwen-14BFP16 · full gpu
DeepSeek-R1-Distill-Qwen-32Bnative BF16 · full gpu
DeepSeek-R1-Distill-Qwen-7BFP16 · full gpu
DeepSeek-V3AWQ · full gpu
Gemma 3 12B ITFP16 · full gpu
Gemma 3 1B ITnative BF16 · full gpu
Gemma 3 27B ITnative BF16 · full gpu
Gemma 3 4B ITFP16 · full gpu
GLM-4.7-Flashnative BF16/F32 · full gpu
gpt-oss-puzzle-88Bnative mxfp4 · full gpu
Ling-3.0-flashQ5_K_M · layer split across gpus
Llama 2 7Bnative F32/F16 · full gpu
Llama 3.1 70B InstructQ8_0 · layer split across gpus
Llama 3.1 8B Instructnative BF16 · full gpu
Llama 3.2 1B InstructFP16 · full gpu
Llama 3.2 3B InstructFP16 · full gpu
Nemotron-Elastic-12Bnative BF16 · full gpu
NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16native BF16 · full gpu
Qwen2.5-14B-InstructFP16 · full gpu
Qwen2.5-32B-InstructFP16 · full gpu
Qwen2.5-72B-InstructQ8_0 · layer split across gpus
Qwen2.5-7B-InstructFP16 · full gpu
Qwen2.5-Coder-32B-InstructFP16 · full gpu
Qwen2.5-Coder-7B-InstructFP16 · full gpu
Step-3.7-FlashQ4_K_M · layer split across gpus
tiny-aya-earthFP16 · full gpu
tiny-aya-fireFP16 · full gpu
tiny-aya-waterFP16 · full gpu
Youtu-Parsing-Omninative BF16 · full gpu

Speed unknown 57

No measurement, calibration or bandwidth figure for this plan.

1
Qwen3.8-Flash-Next
Qwen/Qwen3.8-Flash-Next
2026-08-26180B MoE · 6B active
AA index39.8single mode
Plan & memory
Q4_K_Mllama.cpp ROCmLayer split across GPUsFits*
2 GPUs
115 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
2
Qwen3.8-27B
Qwen/Qwen3.8-27B
2026-08-1427B dense
AA index33.7Xhigh
Plan & memory
native BF16vLLMFull GPUFits*
GPU
53.7 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
3
NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16Specialist
nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16
2026-10-15120B MoE · 12B active
AA index26.5single mode
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsFits*
2 GPUs
123 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
4
Inkling-Small
thinkingmachines/Inkling-Small
2026-07-30276B MoE · 12B active
AA index25.7single mode
Plan & memory
Q4_K_Mllama.cpp ROCmGPU + CPU layers · 34/42 layers on GPUFits
2 GPUs
127 / 128 GiB
RAM
29.1 / 56.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
5
K2-Horizon-MoVA-36B-A4B
IFM/K2-Horizon-MoVA-36B-A4B
2026-09-0336B MoE · 4B active
AA index25.3single mode
Plan & memory
native BF16vLLMTensor split across GPUsFits
2 GPUs
74.2 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
6
Hy3
tencent/Hy3
2026-07-06295B MoE · 21B active
AA index25.3single mode
Plan & memory
Q4_K_Mllama.cpp ROCmGPU + CPU experts · 23 expert layers on CPUFits
2 GPUs
127 / 128 GiB
RAM
48.0 / 56.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
7
G9v3-39A5B (preview)
ai9stars/G9v3-39A5B
2026-08 (day unknown)39B MoE · 5B active
AA index21.8 est.single mode
Plan & memory
native F32/BF16vLLMTensor split across GPUsFits
2 GPUs
75.9 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
8
NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16
nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-BF16
2026-10-1574B MoE · 9B active
AA index21.4single mode
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsFits*
2 GPUs
76.5 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
9
K2-Horizon-7B
IFM/K2-Horizon-7B
2026-09-039B tensors · dense
AA index20.6single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
19.4 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
10
Qwen3.6-35B-A3B
Qwen/Qwen3.6-35B-A3B
2026-04-1635B MoE · 3B active
AA index18.2Reasoning
Plan & memory
native BF16vLLMTensor split across GPUsFits*
2 GPUs
70.1 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
11
Muse-Glimmer-30B
meta-models/Muse-Glimmer-30B
2026-08-1030B dense
AA index17.5High
Plan & memory
native BF16vLLMFull GPUFits
GPU
57.1 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
12
K2-Horizon-3.7B
IFM/K2-Horizon-3.7B
2026-09-035.1B tensors · dense
AA index15.6single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
12.0 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
13
Qwen3.5-122B-A10B
Qwen/Qwen3.5-122B-A10B
2026-02-24122B MoE · 10B active
AA index15.6Reasoning
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsFits*
2 GPUs
127 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
14
Mistral-Medium-3.5-128B
mistralai/Mistral-Medium-3.5-128B
2026-04-28128B dense
AA index14.2single mode
Plan & memory
Q5_K_Mllama.cpp ROCmLayer split across GPUsFits
2 GPUs
88.0 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
15
EXAONE-4.5-33B
LGAI-EXAONE/EXAONE-4.5-33B
2026-04-0933B dense
AA index13.2 est.Reasoning
Plan & memory
native BF16vLLMTensor split across GPUsFits
2 GPUs
68.2 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
16
Qwen3.5-4B
Qwen/Qwen3.5-4B
2026-03-024B dense
AA index13.1 est.Reasoning
Plan & memory
native BF16/F32vLLMFull GPUFits*
GPU
10.4 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
17
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16
nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16
2026-08-1130B MoE · 3B active
AA index12.9single mode
Plan & memory
native F32/BF16vLLMFull GPUFits*
GPU
62.9 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
18
granite-4.2-30b
ibm-granite/granite-4.2-30b
2026-08-2530B dense
AA index12.7single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
58.0 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
19
MiniCPM5-2B
openbmb/MiniCPM5-2B
2026-09-072.5B dense
AA index12.5single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits
GPU
6.5 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
20
Hypernova-60B-2605
MultiverseComputingCAI/Hypernova-60B-2605
2026-05-0660B MoE · 4.8B active
Also: HyperNova-60B (listed with its dated update); Hypernova-60B-2602 (dated update, not separately evaluated)
AA index11.7 est.High
Plan & memory
Q8_0llama.cpp ROCmFull GPUFits
GPU
59.9 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
21
Nemotron-Cascade-2-30B-A3B
nvidia/Nemotron-Cascade-2-30B-A3B
2026-03-1930B MoE · 3B active
AA index11.7 est.single mode
Plan & memory
native F32/BF16vLLMFull GPUFits*
GPU
60.4 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
22
gpt-oss-120bOlder
openai/gpt-oss-120b
2025-08-05117B tensors · MoE
AA index11.6High
Plan & memory
Q8_0llama.cpp ROCmFull GPUFits
GPU
60.8 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
23
Qwen3.5-9B
Qwen/Qwen3.5-9B
2026-03-029B dense
AA index11.2Reasoning
Plan & memory
native BF16/F32vLLMFull GPUFits*
GPU
19.7 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
24
granite-4.2-8b
ibm-granite/granite-4.2-8b
2026-08-258B dense
AA index11.1single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
19.1 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
25
G9v3-3B
ai9stars/G9v3-3B
2026-07 (day unknown)3B dense
AA index10.8 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits
GPU
7.5 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
26
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16Specialist
nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16
2026-04-2830B MoE · 3B active
AA index10.3 est.single mode
Plan & memory
native F32/BF16vLLMFull GPUFits*
GPU
63.0 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
27
North-Mini-Code-1.0Specialist
CohereLabs/North-Mini-Code-1.0
2026-06-0930B MoE · 3B active
AA index9.9single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
58.8 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
28
granite-4.2-3b
ibm-granite/granite-4.2-3b
2026-08-253B dense
AA index9.1single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
8.9 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
29
Tri-21B-Think
trillionlabs/Tri-21B-Think
2026-02 (day unknown)21B dense
AA index9.0 est.single mode
Plan & memory
native F32vLLMTensor split across GPUsFits
2 GPUs
81.8 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
30
gpt-oss-20bOlder
openai/gpt-oss-20b
2025-08-0521B tensors · MoE
AA index9.0High
Plan & memory
native mxfp4vLLMFull GPUFits
GPU
14.5 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
31
NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
2025-12-1532B tensors · MoE
AA index8.9Reasoning
Plan & memory
native F32/BF16vLLMFull GPUFits*
GPU
60.4 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
32
MiniCPM5-1B
openbmb/MiniCPM5-1B
2026-05-191.1B dense
AA index8.8 est.Reasoning
Plan & memory
native BF16vLLMFull GPUFits
GPU
3.7 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
33
Nanbeige4.1-3B
Nanbeige/Nanbeige4.1-3B
2026-02 (day unknown)3B dense
AA index8.4 est.single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
9.3 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
34
LFM2.5-2.6B
LiquidAI/LFM2.5-2.6B
2026-08-042.7B dense
AA index8.4 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits*
GPU
6.7 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
35
Step3-VL-10B
stepfun-ai/Step3-VL-10B
2026-01 (day unknown)10B dense
AA index7.7 est.single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
21.6 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
36
NVIDIA-Nemotron-Nano-12B-v2-VL-BF16Specialist
nvidia/NVIDIA-Nemotron-Nano-12B-v2-VL-BF16
2025-10-2813B tensors · dense
AA index7.5 est.Reasoning
Plan & memory
native F32/BF16vLLMFull GPUFits*
GPU
26.2 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
37
LFM2.5-8B-A1B
LiquidAI/LFM2.5-8B-A1B
2026-05-288.3B MoE · 1.5B active
AA index7.2 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits*
GPU
17.4 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
38
Olmo-3.1-32B-ThinkSpecialist
allenai/Olmo-3.1-32B-Think
2025-12-1232B tensors · dense
AA index7.1 est.single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
62.8 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
39
Qwen3.5-2B
Qwen/Qwen3.5-2B
2026-03-022B dense
AA index6.9 est.Reasoning
Plan & memory
native F32/BF16vLLMFull GPUFits*
GPU
5.8 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
40
Olmo-3.1-32B-Instruct
allenai/Olmo-3.1-32B-Instruct
2025-12-1232B tensors · dense
AA index6.5 est.single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
62.8 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
41
Qwen3.5-0.8B
Qwen/Qwen3.5-0.8B
2026-03-020.8B dense
AA index6.1 est.Reasoning
Plan & memory
native F32/BF16vLLMFull GPUFits*
GPU
3.2 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
42
granite-4.0-h-small
ibm-granite/granite-4.0-h-small
2025-10-0232B tensors · MoE
AA index6.0 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits*
GPU
61.6 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
43
Phi-4Older
microsoft/phi-4
2024-12-1215B tensors · dense
AA index5.9 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits
GPU
30.4 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
44
Olmo-3-7B-ThinkSpecialist
allenai/Olmo-3-7B-Think
2025-11-207.3B tensors · dense
AA index5.6 est.single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
17.6 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
45
Olmo-3-7B-Instruct
allenai/Olmo-3-7B-Instruct
2025-11-207.3B tensors · dense
AA index5.2 est.single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
17.6 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
46
LFM2.5-1.2B-ThinkingSpecialist
LiquidAI/LFM2.5-1.2B-Thinking
2026-01-201.2B tensors · dense
AA index5.2 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits*
GPU
3.8 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
47
LFM2.5-1.2B-Instruct
LiquidAI/LFM2.5-1.2B-Instruct
2026-01-051.2B tensors · dense
AA index5.2 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits*
GPU
3.8 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
48
granite-4.0-h-1b
ibm-granite/granite-4.0-h-1b
2025-10-281.5B tensors · MoE
AA index5.2 est.single mode
Plan & memory
native BF16vLLMFull GPUFits*
GPU
4.3 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
49
Apertus-70B-Instruct-2509
swiss-ai/Apertus-70B-Instruct-2509
2025-09-0271B tensors · dense
AA index5.1 est.single mode
Plan & memory
Q8_0llama.cpp ROCmLayer split across GPUsFits
2 GPUs
75.4 / 128 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
50
granite-4.0-micro
ibm-granite/granite-4.0-micro
2025-10-023.4B tensors · MoE
AA index5.1 est.single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
8.5 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
51
Molmo2-8BSpecialist
allenai/Molmo2-8B
2025-12-118.7B tensors · dense
AA index5.0 est.single mode
Plan & memory
native F32vLLMFull GPUFits
GPU
34.9 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
52
Apertus-8B-Instruct-2509
swiss-ai/Apertus-8B-Instruct-2509
2025-09-028.1B tensors · dense
AA index4.8 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits
GPU
17.5 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
53
granite-4.0-350m
ibm-granite/granite-4.0-350m
2025-10-280.4B tensors · MoE
AA index4.8 est.single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
2.4 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
54
granite-4.0-h-350m
ibm-granite/granite-4.0-h-350m
2025-10-280.3B tensors · MoE
AA index4.8 est.single mode
Plan & memory
native BF16vLLMFull GPUFits*
GPU
2.2 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
55
LFM2.5-VL-1.6BSpecialist
LiquidAI/LFM2.5-VL-1.6B
2026-01-051.6B tensors · dense
AA index4.8 est.single mode
Plan & memory
FP16llama.cpp ROCmFull GPUFits*
GPU
3.8 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
56
NVIDIA-Nemotron-3-Nano-4B-BF16
nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16
2026-03-164B dense
AA index4.8single mode
Plan & memory
native BF16vLLMFull GPUFits*
GPU
9.0 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown
57
K2-Horizon-0.9B
IFM/K2-Horizon-0.9B
2026-09-031.1B tensors · dense
AA index3.0single mode
Plan & memory
native BF16vLLMFull GPUFits
GPU
3.9 / 64.0 GiB
Generate—Unknownspeed unknown
First token · 4K prompt—Unknown

+ 76 models without a AA index score

AgentCPM-ExploreFP16 · full gpu
AgentCPM-Reportnative BF16 · full gpu
Agents-A1native BF16 · tensor split across gpus
Agents-A1-4Bnative BF16 · full gpu
DeepSeek-R1-Distill-Llama-70BQ8_0 · layer split across gpus
DeepSeek-R1-Distill-Llama-8BFP16 · full gpu
FrogBoss-32B-2510FP16 · layer split across gpus
FrogMini-14B-2510FP16 · full gpu
FrogNano-4B-2609native BF16 · full gpu
GLM-4.5-AirQ8_0 · layer split across gpus
granite-4.0-1bnative BF16 · full gpu
granite-4.0-h-micronative BF16 · full gpu
granite-4.0-h-tinynative BF16 · full gpu
granite-4.1-30bnative BF16 · full gpu
granite-4.1-3bnative BF16 · full gpu
granite-4.1-8bnative BF16 · full gpu
IQuest-Coder-V1-14B-BaseFP16 · full gpu
IQuest-Coder-V1-14B-Base-Stage1FP16 · full gpu
IQuest-Coder-V1-14B-InstructFP16 · full gpu
IQuest-Coder-V1-14B-ThinkingFP16 · full gpu
IQuest-Coder-V1-40B-Loop-Thinkingnative BF16 · tensor split across gpus
IQuest-Coder-V1-40B-Thinkingnative BF16 · tensor split across gpus
IQuest-Coder-V1-7B-BaseFP16 · full gpu
IQuest-Coder-V1-7B-Base-Stage1FP16 · full gpu
IQuest-Coder-V1-7B-InstructFP16 · full gpu
IQuest-Coder-V1-7B-ThinkingFP16 · full gpu
Jais-3-0.9Bnative BF16 · full gpu
Jais-3-3.7Bnative BF16 · full gpu
Jais-3-32Bnative BF16 · tensor split across gpus
Jais-3-7Bnative BF16 · full gpu
Jais-3-MoVA-36B-A4BQ8_0 · full gpu
K2-Horizon-32Bnative BF16 · tensor split across gpus
Kolibri-1Q8_0 · layer split across gpus
Laguna-S-2.1Q8_0 · layer split across gpus
LFM2.5-1.2B-JPFP16 · full gpu
LFM2.5-230MFP16 · full gpu
LFM2.5-24B-A2BFP16 · full gpu
LFM2.5-350MFP16 · full gpu
LittleLambFP16 · full gpu
LittleLamb-ToolCallingFP16 · full gpu
MagenticBrainnative F32 · full gpu
Mellum2.1-12B-A2.5B-Thinkingnative BF16 · full gpu
MiMo-V2.6-Distill-Qwen-9Bnative BF16 · full gpu
MiniCPM-SALAnative BF16 · full gpu
MiniCPM-V-4.7-35B-A3BFP16 · layer split across gpus
MiniMax-M2.7Q5_K_M · gpu + cpu experts
Mistral 7B Instruct v0.3FP16 · full gpu
Mistral NeMo 12B Instruct (2407)FP16 · full gpu
Mistral Small 3.2 24B Instruct (2506)native BF16 · full gpu
Mixtral 8x22B Instruct v0.1AWQ · tensor split across gpus
Mixtral 8x7B Instruct v0.1native BF16 · tensor split across gpus
Nemotron-3-Super-120B-A12B-BF16-MTPv2native F32/BF16 · full gpu
Nemotron-Cascade-8BFP16 · full gpu
NVIDIA-Nemotron-Labs-3-Elastic-30B-A3B-BF16native F32/BF16 · full gpu
NVIDIA-Nemotron-Nano-9B-v2-JapaneseFP16 · full gpu
Olmo-3-32B-Thinknative BF16 · full gpu
Ornith-1.5-9Bnative BF16 · full gpu
PaCoRe-8Bnative BF16 · full gpu
Penguin-VL-2Bnative BF16 · full gpu
Penguin-VL-8Bnative BF16 · full gpu
Qwen3-0.6BFP16 · full gpu
Qwen3-1.7BFP16 · full gpu
Qwen3-14BFP16 · full gpu
Qwen3-235B-A22BGPTQ · tensor split across gpus
Qwen3-30B-A3Bnative BF16 · full gpu
Qwen3-32Bnative BF16 · tensor split across gpus
Qwen3-4Bnative BF16 · full gpu
Qwen3-8Bnative BF16 · full gpu
Qwen3-Coder-30B-A3B-Instructnative BF16 · full gpu
Qwen3-Nemotron-32B-RLBFFnative BF16 · tensor split across gpus
Qwen3-VL-235B-A22B-InstructQ4_K_M · gpu + cpu experts
Qwen3-VL-235B-A22B-ThinkingQ4_K_M · gpu + cpu experts
Qwen3.5-27Bnative BF16/F32 · full gpu
Qwen3.5-35B-A3Bnative BF16/F32 · tensor split across gpus
Qwen3.6-27Bnative BF16 · full gpu
RLVR-8B-0926FP16 · full gpu

56 models with no feasible plan on this setup

MiMo-V2.6-Pro-MOPDnative fp8 527 GiB exceeds memory
GLM-5.3native fp8 704 GiB exceeds memory
Step-5-Previewnative fp8 600 GiB exceeds memory
Kimi-K3native F32/BF16 1,454 GiB exceeds memory
GLM-5.3-Flashnative fp8 306 GiB exceeds memory
Ling-3.1-flashQ4_K_M 333 GiB exceeds memory
Qwen3.8-2.4T-A95Bnative BF16 4,556 GiB exceeds memory
DeepSeek-V4.1-FlashQ4_K_M 414 GiB exceeds memory
Muse-Spark-Open-816B-A42BQ4_K_M 465 GiB exceeds memory
MiMo-V2.6-Flash-MOPDnative fp8 161 GiB exceeds memory
DeepSeek-V4-Pro-0813native fp8 831 GiB exceeds memory
Motif-3native BF16 586 GiB exceeds memory
K2-Horizon-375B-A23Bnative BF16 706 GiB exceeds memory
Beam-1.0-501B-A23BQ5_K_M 331 GiB exceeds memory
MiniMax-M3Q4_K_M 247 GiB exceeds memory
Nex-N2-ProQ4_K_M 225 GiB exceeds memory
Kimi-K2.7-Codenative BF16/F32 554 GiB exceeds memory
InklingQ8_0 798 GiB exceeds memory
Solar-Open2-250Bnative BF16/F32 466 GiB exceeds memory
NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16native BF16/F32 1,044 GiB exceeds memory
A.X-K2native fp8 647 GiB exceeds memory
K-EXAONE-2.0-750B-A37Bnative F32/BF16 1,396 GiB exceeds memory
LongCat-2.0native BF16/F32 3,308 GiB exceeds memory
Qwen3.5-397B-A17BGPTQ 220 GiB exceeds memory
Ring-2.6-1Tnative quantized 970 GiB exceeds memory
command-a-plus-05-2026-bf16native BF16 407 GiB exceeds memory
Qwen3-Next-80B-A3B-Thinkingnative BF16 151 GiB exceeds memory
Trinity-Large-ThinkingQ4_K_M 225 GiB exceeds memory
Llama 4 Maverick 17B-128E InstructQ4_K_M 226 GiB exceeds memory
sarvam-105bnative F32 395 GiB exceeds memory
Agnes-2.5-Pro-Alphanative BF16 739 GiB exceeds memory
Agnes-3.0-Qwennative BF16/I64 335 GiB exceeds memory
Apertus-v1.1-0.5B-InstructContext 8K exceeds this model's configured maximum (4K)
Apertus-v1.1-1.5B-InstructContext 8K exceeds this model's configured maximum (4K)
Apertus-v1.1-4B-InstructContext 8K exceeds this model's configured maximum (4K)
Apertus-v1.5-8Bno backend/format combination
BaguettotronContext 8K exceeds this model's configured maximum (4K)
DeepSeek-R1Q4_K_M 377 GiB exceeds memory
DeepSeek-V4-Flash-0731native fp8 155 GiB exceeds memory
GLM-5.2Q4_K_M 434 GiB exceeds memory
Hy3-previewnative BF16/F32 557 GiB exceeds memory
Hy4-previewnative BF16/F32 1,453 GiB exceeds memory
Jais-3-375B-A23Bnative BF16/F32 706 GiB exceeds memory
Kimi-K2.6native F32/I32 554 GiB exceeds memory
Ming-flash-omni-2.0native BF16 194 GiB exceeds memory
Ming-flash-omni-Previewnative BF16/U8 194 GiB exceeds memory
MonadContext 8K exceeds this model's configured maximum (2K)
NVIDIA-Nemotron-3-Super-120B-A12B-BF16native F32/BF16 230 GiB exceeds memory
NVIDIA-Nemotron-Labs-3-Competitive-Coding-550B-A55B-NVFP4native modelopt 328 GiB exceeds memory
NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-BF16native BF16/F32 146 GiB exceeds memory
NVIDIA-Nemotron-Labs-Teacher-Chatnative F32/BF16 1,044 GiB exceeds memory
NVIDIA-Nemotron-Labs-Teacher-Competition-Codingnative BF16/F32 1,044 GiB exceeds memory
NVIDIA-Nemotron-Labs-Teacher-Instruction-Followingnative BF16/F32 1,044 GiB exceeds memory
NVIDIA-Nemotron-Labs-Teacher-STEMnative BF16/F32 1,044 GiB exceeds memory
Qwen3-Coder-480B-A35B-InstructQ4_K_M 270 GiB exceeds memory
Apertus-v1.5-70Bno verified artifact for these backends

Specs & sources

Memory	128 GB HBM2e	source
Bandwidth	3,200 GB/s	source
FP16	362.1 TFLOPS	source
Power	560 W	source

FP16: dense_matrix_fp16. Power: module peak TDP;500W standard/560W peak.

Count this as one physical OAM module containing two logical GPU devices. Backend placement must model each local HBM tier and fabric traffic.

Reported speeds

0 observations

No reported observation resolved to this exact device yet.

Open in speed database Submit yours

Doesn't fit? Rent a GPU

RunpodPlaceholderCloud GPU pods you can rent for running larger models.
Vast.aiPlaceholderGPU rental marketplace with many hosts and GPU types.
LambdaPlaceholderCloud GPU instances for AI training and inference.
TensorDockPlaceholderGPU cloud marketplace with consumer and datacenter GPUs.

Sponsored placeholder links: LocalFit may earn a commission if these become affiliate links. They never influence fit results, estimates or picks.