Gemma 2 9B IT vs Gemma 3 4B IT
Gemma 2 9B IT leads llmrun Score 68.0 to 64.0 (ranges overlap); both fit 24 GB at Q4_K_M; Gemma 3 4B IT is 2.2× faster on RTX 4090 (~231 vs ~107 tok/s).
Choose Gemma 2 9B IT if you want…
Higher llmrun Score
68.0 vs 64.0, ranges overlap
Also does chat
Gemma 3 4B IT doesn't list it
Choose Gemma 3 4B IT if you want…
2.2× faster on RTX 4090
~231 vs ~107 tok/s
Lighter on memory
2.8 vs 6.1 GB at Q4_K_M
Also does vision
Gemma 2 9B IT doesn't list it
Newer release
Feb 2025 vs Jun 2024
On your hardware
Each model at its default quant (Q4_K_M), a smaller one when it must. Estimates, ±20%.
8 GB
RTX 4060 · 8 GB tier
Gemma 2 9B IT
Q4_K_M · ~29 tok/s
7.3 GB · up to 8K
Gemma 3 4B IT
Q4_K_M · ~62 tok/s
5 GB · up to 32Kfaster
12 GB
RTX 3060 12GB · 12 GB tier
Gemma 2 9B IT
Q4_K_M · ~38 tok/s
7.3 GB · up to 8K
Gemma 3 4B IT
Q4_K_M · ~82 tok/s
5 GB · up to 32Kfaster
16 GB
RTX 5060 Ti 16GB · 16 GB tier
Gemma 2 9B IT
Q4_K_M · ~48 tok/s
7.3 GB · up to 8K
Gemma 3 4B IT
Q4_K_M · ~103 tok/s
5 GB · up to 32Kfaster
24 GBreference
RTX 4090 · 24 GB tier
Gemma 2 9B IT
Q4_K_M · ~107 tok/s
7.3 GB · up to 8K
Gemma 3 4B IT
Q4_K_M · ~231 tok/s
5 GB · up to 32Kfaster
32 GB
RTX 5090 · 32 GB tier
Gemma 2 9B IT
Q4_K_M · ~191 tok/s
7.3 GB · up to 8K
Gemma 3 4B IT
Q4_K_M · ~410 tok/s
5 GB · up to 32Kfaster
48 GB
RTX A6000 · 48 GB tier
Gemma 2 9B IT
Q4_K_M · ~82 tok/s
7.3 GB · up to 8K
Gemma 3 4B IT
Q4_K_M · ~176 tok/s
5 GB · up to 32Kfaster
Mac 32 GB
Mac Mini M4 · Macs
Gemma 2 9B IT
Q4_K_M · ~14 tok/s
7.3 GB · up to 8K
Gemma 3 4B IT
Q4_K_M · ~30 tok/s
5 GB · up to 32Kfaster
Mac 64 GB
Mac Studio M4 Max · Macs
Gemma 2 9B IT
Q4_K_M · ~63 tok/s
7.3 GB · up to 8K
Gemma 3 4B IT
Q4_K_M · ~135 tok/s
5 GB · up to 32Kfaster
Mac 128 GB
Mac Studio M4 Max · Macs
Gemma 2 9B IT
Q4_K_M · ~63 tok/s
7.3 GB · up to 8K
Gemma 3 4B IT
Q4_K_M · ~135 tok/s
5 GB · up to 32Kfaster
Reference GPU: RTX 4090. Tinted cell = the better pick on that machine (fits, more headroom, or at least 10% faster).
Overall
Gemma 2 9B IT
Gemma 3 4B IT
Reasoning
Gemma 2 9B IT
no data
Gemma 3 4B IT
Gemma 2 9B IT · #93 of 100 open models · #29 of 35 that fit 24 GB
Gemma 3 4B IT · #97 of 100 open models · #33 of 35 that fit 24 GB
Benchmarks
3 shared boards: Gemma 2 9B IT leads 2, Gemma 3 4B IT leads 1 · Benchmarks imported Oct 2026
Gemma 2 9B IT
27.5%#86/98 open
IndependentEpoch AI · Jun 2024
Gemma 3 4B IT
23.2%#93/98 open
IndependentEpoch AI · Mar 2025
Gemma 2 9B IT
0.6%#82/86 open
IndependentEpoch AI · Jun 2024
Gemma 3 4B IT
7.5%#64/86 open
IndependentEpoch AI · Mar 2025
Gemma 2 9B IT
52.1%#73/145 open
IndependentTIGER-Lab
Gemma 3 4B IT
43.6%#92/145 open
IndependentTIGER-Lab
5 more boards (other shared and single-model boards)
Gemma 2 9B IT
1267.0#124/187 open
ArenaLMArena · Oct 2026
Gemma 3 4B IT
1303.3#105/187 open
ArenaLMArena · Oct 2026
Only Gemma 2 9B IT (2)
- MATH Level 521.0%
- MMLU72.1%
VRAM by quantization
Memory to load the model and 2K context; below it, the total at 32K and 128K context. Lighter side in bold.
Q2_K
Gemma 2 9B IT
4.3 GB3.8 GB file
32K — · 128K —
Gemma 3 4B IT
2.0 GB1.7 GB file
32K 6.6 GB · 128K 21.4 GB
Q3_K_S
Gemma 2 9B IT
4.5 GB4.3 GB file
32K — · 128K —
Gemma 3 4B IT
2.1 GB1.9 GB file
32K 6.7 GB · 128K 21.5 GB
Q3_K_M
Gemma 2 9B IT
5.0 GB4.8 GB file
32K — · 128K —
Gemma 3 4B IT
2.3 GB2.1 GB file
32K 6.9 GB · 128K 21.7 GB
Q4_0
Gemma 2 9B IT
5.1 GB5.4 GB file
32K — · 128K —
Gemma 3 4B IT
2.4 GB2.4 GB file
32K 7.0 GB · 128K 21.8 GB
Q4_K_M
Gemma 2 9B IT
6.1 GB5.8 GB file
32K — · 128K —
Gemma 3 4B IT
2.8 GB2.5 GB file
32K 7.5 GB · 128K 22.3 GB
Q5_K_M
Gemma 2 9B IT
7.2 GB6.6 GB file
32K — · 128K —
Gemma 3 4B IT
3.4 GB2.8 GB file
32K 8.0 GB · 128K 22.8 GB
Q6_K
Gemma 2 9B IT
8.4 GB7.6 GB file
32K — · 128K —
Gemma 3 4B IT
3.9 GB3.2 GB file
32K 8.5 GB · 128K 23.3 GB
Q8_0
Gemma 2 9B IT
10.2 GB9.8 GB file
32K — · 128K —
Gemma 3 4B IT
4.7 GB4.1 GB file
32K 9.4 GB · 128K 24.1 GB
20 more quants
IQ2_XXS
Gemma 2 9B IT
2.8 GB2.8 GB file
32K — · 128K —
Gemma 3 4B IT
1.3 GB1.3 GB file
32K 5.9 GB · 128K 20.7 GB
IQ2_XS
Gemma 2 9B IT
3.1 GB3.1 GB file
32K — · 128K —
Gemma 3 4B IT
1.4 GB1.4 GB file
32K 6.0 GB · 128K 20.8 GB
IQ2_S
Gemma 2 9B IT
3.2 GB3.2 GB file
32K — · 128K —
Gemma 3 4B IT
1.5 GB1.4 GB file
32K 6.1 GB · 128K 20.9 GB
IQ2_M
Gemma 2 9B IT
3.4 GB3.4 GB file
32K — · 128K —
Gemma 3 4B IT
1.6 GB1.6 GB file
32K 6.2 GB · 128K 21.0 GB
IQ3_XXS
Gemma 2 9B IT
3.9 GB3.8 GB file
32K — · 128K —
Gemma 3 4B IT
1.8 GB1.7 GB file
32K 6.5 GB · 128K 21.2 GB
Q2_K_S
Gemma 2 9B IT
not listed
Gemma 3 4B IT
1.9 GB1.6 GB file
32K 6.5 GB · 128K 21.3 GB
IQ3_XS
Gemma 2 9B IT
4.2 GB4.1 GB file
32K — · 128K —
Gemma 3 4B IT
2.0 GB1.9 GB file
32K 6.6 GB · 128K 21.4 GB
IQ3_S
Gemma 2 9B IT
4.3 GB4.3 GB file
32K — · 128K —
Gemma 3 4B IT
2.0 GB1.9 GB file
32K 6.6 GB · 128K 21.4 GB
IQ3_M
Gemma 2 9B IT
4.6 GB4.5 GB file
32K — · 128K —
Gemma 3 4B IT
2.1 GB2.0 GB file
32K 6.8 GB · 128K 21.5 GB
Q3_K_L
Gemma 2 9B IT
5.2 GB5.1 GB file
32K — · 128K —
Gemma 3 4B IT
2.4 GB2.2 GB file
32K 7.0 GB · 128K 21.8 GB
IQ4_XS
Gemma 2 9B IT
5.5 GB5.2 GB file
32K — · 128K —
Gemma 3 4B IT
2.5 GB2.3 GB file
32K 7.2 GB · 128K 22.0 GB
IQ4_NL
Gemma 2 9B IT
not listed
Gemma 3 4B IT
2.7 GB2.4 GB file
32K 7.3 GB · 128K 22.1 GB
Q4_1
Gemma 2 9B IT
5.7 GB6.0 GB file
32K — · 128K —
Gemma 3 4B IT
2.7 GB2.6 GB file
32K 7.3 GB · 128K 22.1 GB
Q4_K_S
Gemma 2 9B IT
5.7 GB5.5 GB file
32K — · 128K —
Gemma 3 4B IT
2.7 GB2.4 GB file
32K 7.3 GB · 128K 22.1 GB
Q4_K_L
Gemma 2 9B IT
6.2 GB6.0 GB file
32K — · 128K —
Gemma 3 4B IT
2.9 GB2.7 GB file
32K 7.5 GB · 128K 22.3 GB
Q5_0
Gemma 2 9B IT
6.4 GB6.5 GB file
32K — · 128K —
Gemma 3 4B IT
3.0 GB2.8 GB file
32K 7.6 GB · 128K 22.4 GB
Q5_1
Gemma 2 9B IT
7.0 GB7.0 GB file
32K — · 128K —
Gemma 3 4B IT
not listed
Q5_K_S
Gemma 2 9B IT
7.0 GB6.5 GB file
32K — · 128K —
Gemma 3 4B IT
3.3 GB2.8 GB file
32K 7.9 GB · 128K 22.7 GB
Q5_K_L
Gemma 2 9B IT
7.4 GB6.9 GB file
32K — · 128K —
Gemma 3 4B IT
3.4 GB3.0 GB file
32K 8.1 GB · 128K 22.8 GB
BF16
Gemma 2 9B IT
20.3 GB~18.5 GB file
32K — · 128K —
Gemma 3 4B IT
9.5 GB7.8 GB file
32K 14.1 GB · 128K 28.9 GB
Gemma 2 9B IT
Ollama · Q4_K_M
ollama run hf.co/bartowski/gemma-2-9b-it-GGUF:Q4_K_MGemma 3 4B IT
Ollama · Q4_K_M
ollama run hf.co/MaziyarPanahi/gemma-3-4b-it-GGUF:Q4_K_M| Spec | Gemma 2 9B IT | Gemma 3 4B IT |
|---|---|---|
| Publisher | ||
| Parameters | 9.2B | 4.3B |
| Context | 8K tokens | — |
| Modalities | Chat | Vision |
| License | Gemma Terms | Gemma Terms |
| Released | Jun 2024 | Feb 2025 |
Frequently Asked Questions
Which is better, Gemma 2 9B IT or Gemma 3 4B IT?
Gemma 2 9B IT leads llmrun Score 68.0 to 64.0 (ranges overlap); both fit 24 GB at Q4_K_M; Gemma 3 4B IT is 2.2× faster on RTX 4090 (~231 vs ~107 tok/s). Choose Gemma 2 9B IT for higher llmrun Score and also does chat; choose Gemma 3 4B IT for 2.2× faster on RTX 4090 and lighter on memory.
Which needs less VRAM, Gemma 2 9B IT or Gemma 3 4B IT?
At Q4_K_M, Gemma 2 9B IT requires 6.1 GB while Gemma 3 4B IT needs 2.8 GB, making Gemma 3 4B IT the easier model to run on your local setup.
Which is faster on RTX 4090?
Estimated decode speed on RTX 4090 is about 107 tok/s for Gemma 2 9B IT (Q4_K_M) and 231 tok/s for Gemma 3 4B IT (Q4_K_M), so Gemma 3 4B IT is faster. Estimates, ±20%.
Can I run Gemma 2 9B IT or Gemma 3 4B IT on a 16 GB GPU?
Gemma 2 9B IT fits at Q4_K_M; Gemma 3 4B IT fits at Q4_K_M.