NVIDIA GeForce RTX 4090 vs NVIDIA GeForce RTX 5070

Head-to-head AI inference comparison across 109 popular models. Each model is graded against both cards using its highest-quality quantization that still fits in VRAM. Bigger grade and faster tokens-per-second wins.

Spec
  • VRAM
  • Architecture
  • Vendor
  • MSRP
  • Models running
  • Wins (grade)
NVIDIA GeForce RTX 4090
  • 24GB
  • ada lovelace
  • nvidia
  • $1,599
  • 109 of 109
  • 10 models
NVIDIA GeForce RTX 5070
  • 12GB
  • blackwell
  • nvidia
  • $549
  • 108 of 109
  • 0 models

Where NVIDIA GeForce RTX 4090 pulls ahead

Where NVIDIA GeForce RTX 5070 pulls ahead

  • No standout wins.

Language Models47 models

S
34
Gemma 3 27B
27B · Google
—
D
A
34
Qwen 2.5 32B
32B · Alibaba
—
D
S
34
Mistral Small 22B
22B · Mistral AI
—
C
D
—
Llama 3.1 70B Instruct
70B · Meta
—
F
S
66
Phi-4
14B · Microsoft
36
A
S
66
Qwen 2.5 14B
14B · Alibaba
36
A
S
66
Gemma 3 12B
12B · Google
36
S
S
66
Mistral Nemo 12B
12B · Mistral AI
36
S
S
66
Solar 10.7B
10.7B · Upstage
36
S
S
96
Falcon 3 10B
10B · TII
62
S
S
96
Gemma 2 9B Instruct
9.2B · Google
62
S
S
96
Yi 1.5 9B Chat
9B · 01.AI
62
S
S
96
DeepSeek R1 Distill 8B
8B · DeepSeek
62
S
S
96
Llama 3.1 8B Instruct
8B · Meta
62
S
S
96
Granite 3.3 8B
8B · IBM
62
S
S
96
EXAONE 3.5 7.8B
7.8B · LG AI
62
S
S
96
InternLM 2.5 7B
7.7B · Shanghai AI Lab
62
S
S
96
Qwen 2.5 7B Instruct
7.6B · Alibaba
62
S
S
96
Mistral 7B Instruct v0.3
7.3B · Mistral AI
62
S
S
96
Falcon 3 7B
7B · TII
62
S
S
96
OLMo 2 7B
7B · Allen AI
62
S
S
96
OpenChat 3.5 7B
7B · OpenChat
62
S
S
96
Yi 1.5 6B Chat
6B · 01.AI
62
S
S
144
Gemma 3 4B
4B · Google
94
S
S
144
Nemotron Mini 4B
4B · NVIDIA
94
S
S
144
Danube 3 4B
4B · H2O.ai
94
S
S
144
Phi-3.5 Mini 3.8B
3.8B · Microsoft
94
S
S
144
Phi-4 Mini 3.8B
3.8B · Microsoft
94
S
S
144
Llama 3.2 3B Instruct
3.2B · Meta
94
S
S
144
Qwen 2.5 3B
3B · Alibaba
94
S
S
144
Falcon 3 3B
3B · TII
94
S
S
144
StableLM Zephyr 3B
3B · Stability AI
94
S
S
144
Rocket 3B
3B · Pansophic
94
S
S
144
Gemma 2 2B
2.6B · Google
94
S
S
144
EXAONE 3.5 2.4B
2.4B · LG AI
94
S
S
192
Granite 3.3 2B
2B · IBM
132
S
S
192
SmolLM2 1.7B
1.7B · HuggingFace
132
S
S
192
Qwen 2.5 1.5B
1.5B · Alibaba
132
S
S
192
DeepSeek R1 Distill 1.5B
1.5B · DeepSeek
132
S
S
192
Llama 3.2 1B Instruct
1.24B · Meta
132
S
S
192
TinyLlama 1.1B
1.1B · TinyLlama
132
S
S
192
Gemma 3 1B
1B · Google
132
S
S
192
Falcon 3 1B
1B · TII
132
S
S
192
Qwen 2.5 0.5B
0.5B · Alibaba
132
S
S
192
Danube 3 500M
0.5B · H2O.ai
132
S
S
192
SmolLM2 360M
0.36B · HuggingFace
132
S
S
192
SmolLM2 135M
0.135B · HuggingFace
132
S

Code Models16 models

S
66
Qwen 2.5 Coder 14B
14B · Alibaba
36
A
S
66
Code Llama 13B Instruct
13B · Meta
36
S
S
96
Yi Coder 9B
9B · 01.AI
62
S
S
96
CodeGemma 7B
8.5B · Google
62
S
S
96
Qwen 2.5 Coder 7B
7.6B · Alibaba
62
S
S
96
StarCoder2 7B
7B · BigCode
62
S
S
96
Code Llama 7B
7B · Meta
62
S
S
96
DeepSeek Coder 6.7B
6.7B · DeepSeek
62
S
S
144
Qwen 2.5 Coder 3B
3B · Alibaba
94
S
S
144
StarCoder2 3B
3B · BigCode
94
S
S
144
Stable Code 3B
3B · Stability AI
94
S
S
192
CodeGemma 2B
2B · Google
132
S
S
192
Qwen 2.5 Coder 1.5B
1.5B · Alibaba
132
S
S
192
Yi Coder 1.5B
1.5B · 01.AI
132
S
S
192
DeepSeek Coder 1.3B
1.3B · DeepSeek
132
S
S
192
Qwen 2.5 Coder 0.5B
0.5B · Alibaba
132
S

Multimodal & Vision6 models

S
96
LLaVA 1.6 7B
7B · LLaVA
62
S
S
144
Phi-3.5 Vision
4.2B · Microsoft
94
S
S
144
PaliGemma 3B
3B · Google
94
S
S
144
Qwen2-VL 2B
2.2B · Alibaba
94
S
S
192
MiniCPM-V 2.6
2B · OpenBMB
132
S
S
192
Moondream 2
1.8B · Moondream
132
S

Image Generation9 models

S
66
FLUX.1 Schnell (GGUF)
12B · Black Forest Labs
26
C
S
66
FLUX.1 Dev (GGUF)
12B · Black Forest Labs
26
C
S
144
Stable Diffusion 3 Medium (GGUF)
2.5B · Stability AI
94
A
S
144
Stable Diffusion XL (CoreML)
3.5B · Stability AI
94
S
S
144
SDXL Turbo (GGUF)
3.5B · Stability AI
94
S
S
192
Stable Diffusion 2.1 Base (CoreML)
0.86B · Stability AI / Apple
132
S
S
192
Stable Diffusion 1.5 (CoreML)
0.86B · Runway
132
S
S
192
Stable Diffusion 1.5 (GGUF)
0.86B · Runway / GPUStack
132
S
S
192
Stable Diffusion 2.1 (GGUF)
0.86B · Stability AI
132
S

Speech9 models

S
192
Whisper Large v3
1.55B · OpenAI
132
S
S
192
Whisper Large v3 Turbo
0.81B · OpenAI
132
S
S
192
Whisper Medium
0.77B · OpenAI
132
S
S
192
Distil-Whisper Large v3
0.76B · HuggingFace
132
S
S
192
Whisper Small
0.24B · OpenAI
132
S
S
192
Whisper Base
0.074B · OpenAI
132
S
S
192
Whisper Base English
0.074B · OpenAI
132
S
S
192
Whisper Tiny English (Quantized)
0.039B · OpenAI
132
S
S
192
Whisper Tiny
0.039B · OpenAI
132
S

Text-to-Speech14 models

S
192
Kokoro 82M TTS
0.082B · Kokoro
132
S
S
192
Piper TTS - Amy (English)
0.02B · Rhasspy
132
S
S
192
Piper TTS - Lessac (English)
0.02B · Rhasspy
132
S
S
192
Piper TTS - LibriTTS-R (English)
0.02B · Rhasspy
132
S
S
192
Piper TTS - Spanish (MLS)
0.02B · Rhasspy
132
S
S
192
Piper TTS - French (Siwis)
0.02B · Rhasspy
132
S
S
192
Piper TTS - German (Thorsten)
0.02B · Rhasspy
132
S
S
192
Piper TTS - Chinese (Huayan)
0.02B · Rhasspy
132
S
S
192
Piper TTS - Japanese (Kokoro)
0.02B · Rhasspy
132
S
S
192
Piper TTS - Korean
0.02B · Rhasspy
132
S
S
192
Piper TTS - Russian (Irina)
0.02B · Rhasspy
132
S
S
192
Piper TTS - Portuguese (Faber)
0.02B · Rhasspy
132
S
S
192
Piper TTS - Italian (Riccardo)
0.02B · Rhasspy
132
S
S
192
Piper TTS - Arabic (Kareem)
0.02B · Rhasspy
132
S

Embeddings5 models

S
192
BGE Large EN v1.5
0.335B · BAAI
132
S
S
192
Nomic Embed Text v1.5
0.137B · Nomic AI
132
S
S
192
BGE Small EN v1.5
0.033B · BAAI
132
S
S
192
Snowflake Arctic Embed S
0.033B · Snowflake
132
S
S
192
all-MiniLM-L6-v2
0.023B · Sentence Transformers
132
S

Rerankers2 models

S
192
BGE Reranker v2 M3
0.568B · BAAI
132
S
S
192
Jina Reranker Tiny EN
0.033B · Jina AI
132
S