RUNLOCALAIv38
->Will it run?Best GPUCompareTroubleshootStartLearnPulseModelsHardwareToolsBench
Run check
RUNLOCALAI

Independently operated catalog for local-AI hardware and software. Hand-written verdicts. Source-cited claims. Reproducible commands when we have them.

OP·Eruo Fredoline
DIR
  • Models
  • Hardware
  • Tools
  • Benchmarks
TOOLS
  • Will it run?
  • Compare hardware
  • Cost vs cloud
  • Choose my GPU
  • Prompting kits
  • Quick answers
REF
  • All buyer guides
  • Learn local AI
  • Methodology
  • Glossary
  • Errors KB
  • Trust
EDITOR
  • About
  • Author
  • How we make money
  • Editorial policy
  • Contact
LEGAL
  • Privacy
  • Terms
  • Sitemap
MAIL · MONTHLY DIGEST
Get monthly local AI changes
Monthly recap. No spam.
DISCLOSURE

Some links on this site are affiliate links (Amazon Associates and other first-class retailers). When you buy through them, we earn a small commission at no extra cost to you. Affiliate links do not influence our verdicts — there are cards we rate highly that we don't have affiliate relationships with, and cards that sell well that we refuse to recommend. Read more →

© 2026 runlocalai.coIndependently operated
RUNLOCALAI · v38
Will it run? / NVIDIA GeForce RTX 5090

What can NVIDIA GeForce RTX 5090 run?

Build: RTX 5090 + Ryzen 9 9950X + 64GB DDR5

Memory: 32 GB VRAM + 64 GB system RAM
Runner: llama.cpp / Ollama (CUDA)
AnyChatCodingAgentsReasoningVisionLong contextCreative

Runs comfortably
249 models

Full-VRAM resident, with room for context. No compromises.

#1all-MiniLM-L6-v2
0.022B
other
Commercial OK
Quant: Q4_K_MContext: 256VRAM: 1.8 GBHeadroom: 30.2 GBTTFT: instant
87694
tok/s
Estimated
Weights
0.01 GB
KV cache
0.00 GB
Activations
0.00 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~0 ms (instant)
Model details →
#2Piper
0.025B
other
Commercial OK
Quant: Q4_K_MContext: 0VRAM: 1.8 GBHeadroom: 30.2 GBTTFT: instant
77171
tok/s
Estimated
Weights
0.02 GB
KV cache
0.00 GB
Activations
0.00 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~1 ms (instant)
Model details →
#3Whisper Tiny
0.039B
other
Commercial OK
Quant: Q4_K_MContext: 30VRAM: 1.8 GBHeadroom: 30.2 GBTTFT: instant
49469
tok/s
Estimated
Weights
0.02 GB
KV cache
0.00 GB
Activations
0.00 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~1 ms (instant)
Model details →
#4Whisper Base
0.074B
other
Commercial OK
Quant: Q4_K_MContext: 30VRAM: 1.8 GBHeadroom: 30.2 GBTTFT: instant
26071
tok/s
Estimated
Weights
0.04 GB
KV cache
0.00 GB
Activations
0.00 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~2 ms (instant)
Model details →
#5Kokoro 82M
0.082B
other
Commercial OK
Quant: Q4_K_MContext: 0VRAM: 1.9 GBHeadroom: 30.1 GBTTFT: instant
23528
tok/s
Estimated
Weights
0.05 GB
KV cache
0.00 GB
Activations
0.00 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~2 ms (instant)
Model details →
#6all-mpnet-base-v2
0.109B
other
Commercial OK
Quant: Q4_K_MContext: 384VRAM: 1.9 GBHeadroom: 30.1 GBTTFT: instant
17700
tok/s
Estimated
Weights
0.07 GB
KV cache
0.00 GB
Activations
0.00 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~2 ms (instant)
Model details →
#7paraphrase-multilingual-MiniLM-L12-v2
0.118B
other
Commercial OK
Quant: Q4_K_MContext: 128VRAM: 1.9 GBHeadroom: 30.1 GBTTFT: instant
16350
tok/s
Estimated
Weights
0.07 GB
KV cache
0.00 GB
Activations
0.00 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~2 ms (instant)
Model details →
#8Nomic Embed Text v1.5
0.137B
other
Commercial OK
Quant: Q4_K_MContext: 8,192VRAM: 2.0 GBHeadroom: 30.0 GBTTFT: instant
14082
tok/s
Estimated
Weights
0.08 GB
KV cache
0.07 GB
Activations
0.01 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~3 ms (instant)
Model details →
#9SmolLM2 135M Instruct
0.135B
other
Commercial OK
Quant: Q4_K_MContext: 8,192VRAM: 2.0 GBHeadroom: 30.0 GBTTFT: instant
14291
tok/s
Estimated
Weights
0.08 GB
KV cache
0.07 GB
Activations
0.01 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~3 ms (instant)
Model details →
#10GTE ModernBERT Base
0.149B
other
Commercial OK
Quant: Q4_K_MContext: 8,192VRAM: 2.0 GBHeadroom: 30.0 GBTTFT: instant
12948
tok/s
Estimated
Weights
0.09 GB
KV cache
0.07 GB
Activations
0.01 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~3 ms (instant)
Model details →
#11Whisper Small
0.244B
other
Commercial OK
Quant: Q4_K_MContext: 30VRAM: 2.0 GBHeadroom: 30.0 GBTTFT: instant
7907
tok/s
Estimated
Weights
0.15 GB
KV cache
0.00 GB
Activations
0.01 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~5 ms (instant)
Model details →
#12Gemma 3 270M
0.27B
gemma
Commercial OK
Quant: Q4_K_MContext: 8,192VRAM: 2.1 GBHeadroom: 29.9 GBTTFT: instant
7145
tok/s
Estimated
Weights
0.16 GB
KV cache
0.14 GB
Activations
0.02 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~6 ms (instant)
Model details →

Runs with tradeoffs
44 models

Tight VRAM, partial CPU offload, or context-limited.

Llama 3.3 70B Instruct
70B
llama
Commercial OK
Quant: Q5_K_MContext: 8,192VRAM: 55.0 GBHeadroom: 15.4 GBTTFT: noticeable
  • • Partial CPU offload: ~42% of layers run on CPU
  • • CPU is the bottleneck — upgrading RAM bandwidth helps more than VRAM here
ollama run llama3.3:70b
2
tok/s
Estimated
Weights
48.13 GB
KV cache
2.68 GB
Activations
2.41 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~1434 ms (noticeable)
Model details →
DeepSeek R1 Distill Llama 70B
70B
deepseek
Commercial OK
Quant: Q4_K_MContext: 2,048VRAM: 54.9 GBHeadroom: 15.5 GBTTFT: noticeable
  • • Partial CPU offload: ~42% of layers run on CPU
  • • CPU is the bottleneck — upgrading RAM bandwidth helps more than VRAM here
ollama run deepseek-r1:70b
2
tok/s
Estimated
Weights
42.26 GB
KV cache
8.75 GB
Activations
2.12 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~1434 ms (noticeable)
Model details →
Qwen3.6 35B-A3B
35B
qwen
Commercial OK
Quant: Q4_K_MContext: 2,048VRAM: 28.4 GBHeadroom: 3.6 GBTTFT: noticeable
  • • Tight VRAM fit — only 3.6 GB headroom left for context growth
ollama run qwen3.6:35b-a3b
55
tok/s
Estimated
Weights
21.13 GB
KV cache
4.38 GB
Activations
1.06 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~717 ms (noticeable)
Model details →
Gemma 4 26B MoE
26B
gemma
Commercial OK
Quant: Q4_K_MContext: 8,192VRAM: 31.3 GBHeadroom: 0.7 GBTTFT: noticeable
  • • Tight VRAM fit — only 0.7 GB headroom left for context growth
ollama run gemma4:26b-moe
74
tok/s
Estimated
Weights
15.70 GB
KV cache
13.00 GB
Activations
0.79 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~532 ms (noticeable)
Model details →
Qwen3.5 35B-A3B
35B
qwen
Commercial OK
Quant: Q4_K_MContext: 2,048VRAM: 28.4 GBHeadroom: 3.6 GBTTFT: noticeable
  • • Tight VRAM fit — only 3.6 GB headroom left for context growth
ollama run qwen3.5:35b-a3b
55
tok/s
Estimated
Weights
21.13 GB
KV cache
4.38 GB
Activations
1.06 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~717 ms (noticeable)
Model details →
Mistral Small 3 24B
24B
mistral
Commercial OK
Quant: Q4_K_MContext: 8,192VRAM: 29.0 GBHeadroom: 3.0 GBTTFT: fast
  • • Tight VRAM fit — only 3.0 GB headroom left for context growth
ollama run mistral-small:24b
80
tok/s
Estimated
Weights
14.49 GB
KV cache
12.00 GB
Activations
0.73 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~492 ms (fast)
Model details →
Gemma 4 26B-A4B
26B
gemma
Commercial OK
Quant: Q4_K_MContext: 8,192VRAM: 31.3 GBHeadroom: 0.7 GBTTFT: noticeable
  • • Tight VRAM fit — only 0.7 GB headroom left for context growth
ollama run gemma4:26b-a4b-it-q4_K_M
74
tok/s
Estimated
Weights
15.70 GB
KV cache
13.00 GB
Activations
0.79 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~532 ms (noticeable)
Model details →
Llama 3.1 70B Instruct
70B
llama
Commercial OK
Quant: Q4_K_MContext: 2,048VRAM: 54.9 GBHeadroom: 15.5 GBTTFT: noticeable
  • • Partial CPU offload: ~42% of layers run on CPU
  • • CPU is the bottleneck — upgrading RAM bandwidth helps more than VRAM here
ollama run llama3.1:70b
2
tok/s
Estimated
Weights
42.26 GB
KV cache
8.75 GB
Activations
2.12 GB
Runtime
1.80 GB
Time to first token (prefill, 512-token prompt): ~1434 ms (noticeable)
Model details →

What if you upgraded?

Hypothetical scenarios. We re-ran the compatibility engine for each.

+32 GB system RAM

~$80–150

Doubles your CPU-offload working set. Helps when models don't quite fit in VRAM.

Unlocks: 53 new tradeoff

  • • Llama 4 Scout
  • • Llama 3.3 70B Instruct
  • • DeepSeek R1 Distill Llama 70B
  • • Qwen3.6 35B-A3B
Shop this upgrade↗

Upgrade to NVIDIA A100 40GB

see current pricing

40 GB VRAM (vs your 32 GB) plus a bandwidth jump from ~1792 GB/s to ~1555 GB/s.

Unlocks: 21 new comfortable

  • • Mistral Small 3 24B
  • • Gemma 4 26B MoE
  • • Gemma 4 26B-A4B
  • • Dolphin 3.0 Mistral 24B
Shop this upgrade↗

Add a second NVIDIA GeForce RTX 5090

~$2499

Tensor parallelism splits the model across both cards, effectively doubling VRAM. Bandwidth doesn't double — runs ~1.5× the single-card speed in practice.

Unlocks: 41 new comfortable

  • • Gemma 4 26B MoE
  • • Gemma 4 26B-A4B
  • • Dolphin 3.0 Mistral 24B
  • • Gemma 4 Turkish 26B (4B active)
Shop this upgrade↗

Some links above are affiliate links. We may earn a commission at no extra cost to you. How we make money.

Won't run
top 5 popular models

Need more memory than you have. Shown for orientation.

DeepSeek V4 Pro (1.6T MoE)
1600B
deepseek
Commercial OK

Even with CPU offload, needs more memory than your VRAM (32 GB) + 60% of system RAM (38 GB) combined.

—
Qwen 3.5 235B-A17B (MoE)
397B
qwen
Commercial OK

Even with CPU offload, needs more memory than your VRAM (32 GB) + 60% of system RAM (38 GB) combined.

—
Qwen 3 235B-A22B
235B
qwen
Commercial OK

Even with CPU offload, needs more memory than your VRAM (32 GB) + 60% of system RAM (38 GB) combined.

—
DeepSeek R1 (671B reasoning)
671B
deepseek
Commercial OK

Even with CPU offload, needs more memory than your VRAM (32 GB) + 60% of system RAM (38 GB) combined.

—
Llama 4 Scout
109B
llama
Commercial OK

Even with CPU offload, needs more memory than your VRAM (32 GB) + 60% of system RAM (38 GB) combined.

—

How to read these numbers

Measured here
Measured here - RunLocalAI ran this exact combo on owner hardware with public evidence.

Source-backed
Source-backed / community - a reproduced public source supports the speed, but it is not labeled as owner-measured.

Extrapolated
Extrapolated - predicted from a measured benchmark on similar-bandwidth hardware.

Estimated
Estimated - formula based on VRAM bandwidth and model architecture; not a benchmark row.

RunLocalAI Will-It-Run Framework →

Want a specific benchmark we don't have? Email Contact support and we'll prioritize it.