Gemma 3 12B IT vs GPT OSS 20B
GPT OSS 20B leads llmrun Score 96.4 to 74.2; both fit 24 GB at Q4_K_M; GPT OSS 20B is 2.3× faster on RTX 4090 (~190 vs ~82 tok/s).
Model B · OpenAI
GPT OSS 20B20.9B · 4.2B active · 128K context · Apache 2.0 · Aug 2025
Choose Gemma 3 12B IT if you want…
Lighter on memory
8.0 vs 12.9 GB at Q4_K_M
Also does vision
GPT OSS 20B doesn't list it
Choose GPT OSS 20B if you want…
Higher llmrun Score
96.4 vs 74.2
Stronger at reasoning
91.3 vs 66.4
Stronger at science
93.1 vs 71.6
Ahead on the benchmarks
7 of 7 shared boards
2.3× faster on RTX 4090
~190 vs ~82 tok/s
Longer context window
128K vs 32K tokens
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 3 12B IT
IQ2_M · ~39 tok/s
7.4 GB · up to 16K
GPT OSS 20B
IQ2_M · ~103 tok/s
7.6 GB · up to 32Kfaster
12 GB
RTX 3060 12GB · 12 GB tier
Gemma 3 12B IT
Q4_K_M · ~29 tok/s
10.9 GB · up to 16Kmore headroom
GPT OSS 20B
IQ4_XS · ~89 tok/s
11.8 GB · up to 16K
16 GB
RTX 5060 Ti 16GB · 16 GB tier
Gemma 3 12B IT
Q4_K_M · ~36 tok/s
10.9 GB · up to 32K
GPT OSS 20B
Q4_K_M · ~98 tok/s
13.1 GB · up to 128Kfaster
24 GBreference
RTX 4090 · 24 GB tier
Gemma 3 12B IT
Q4_K_M · ~82 tok/s
10.9 GB · up to 32K
GPT OSS 20B
Q4_K_M · ~190 tok/s
13.1 GB · up to 128Kfaster
32 GB
RTX 5090 · 32 GB tier
Gemma 3 12B IT
Q4_K_M · ~145 tok/s
10.9 GB · up to 32K
GPT OSS 20B
Q4_K_M · ~282 tok/s
13.1 GB · up to 128Kfaster
48 GB
RTX A6000 · 48 GB tier
Gemma 3 12B IT
Q4_K_M · ~62 tok/s
10.9 GB · up to 32K
GPT OSS 20B
Q4_K_M · ~154 tok/s
13.1 GB · up to 128Kfaster
Mac 32 GB
Mac Mini M4 · Macs
Gemma 3 12B IT
Q4_K_M · ~10 tok/s
10.9 GB · up to 32K
GPT OSS 20B
Q4_K_M · ~28 tok/s
13.1 GB · up to 128Kfaster
Mac 64 GB
Mac Studio M4 Max · Macs
Gemma 3 12B IT
Q4_K_M · ~48 tok/s
10.9 GB · up to 32K
GPT OSS 20B
Q4_K_M · ~86 tok/s
13.1 GB · up to 128Kfaster
Mac 128 GB
Mac Studio M4 Max · Macs
Gemma 3 12B IT
Q4_K_M · ~48 tok/s
10.9 GB · up to 32K
GPT OSS 20B
Q4_K_M · ~86 tok/s
13.1 GB · up to 128Kfaster
Reference GPU: RTX 4090. Tinted cell = the better pick on that machine (fits, more headroom, or at least 10% faster).
Overall
Gemma 3 12B IT
GPT OSS 20B
Coding
Gemma 3 12B IT
no data
GPT OSS 20B
Science
Gemma 3 12B IT
GPT OSS 20B
Reasoning
Gemma 3 12B IT
GPT OSS 20B
Gemma 3 12B IT · #96 of 106 open models · #28 of 36 that fit 24 GB
GPT OSS 20B · #69 of 106 open models · #16 of 36 that fit 24 GB
Benchmarks
7 shared boards: Gemma 3 12B IT leads 0, GPT OSS 20B leads 7 · Benchmarks imported Nov 2026
Gemma 3 12B IT
39.5%#73/100 open
IndependentEpoch AI · Mar 2025
GPT OSS 20B
60.8%#44/100 open
IndependentEpoch AI · Aug 2025
Gemma 3 12B IT
16.7%#57/88 open
IndependentEpoch AI · Mar 2025
GPT OSS 20B
65.3%#36/88 open
IndependentEpoch AI · Aug 2025
Gemma 3 12B IT
60.6%#57/146 open
IndependentTIGER-Lab
GPT OSS 20B
73.6%#33/146 open
IndependentTIGER-Lab
Gemma 3 12B IT
16.4%#70/72 open
Lab-reportedEpoch AI · Mar 2025
GPT OSS 20B
38.9%#41/72 open
Lab-reportedEpoch AI · Aug 2025
Gemma 3 12B IT
0.0%#56/84 open
Lab-reportedEpoch AI · Mar 2025
GPT OSS 20B
1.4%#36/84 open
Lab-reportedEpoch AI · Aug 2025
Gemma 3 12B IT
48.8%#64/70 open
Lab-reportedEpoch AI · Mar 2025
GPT OSS 20B
68.0%#36/70 open
Lab-reportedEpoch AI · Aug 2025
Gemma 3 12B IT
4.5%#55/56 open
Lab-reportedEpoch AI · Mar 2025
GPT OSS 20B
14.5%#45/56 open
Lab-reportedEpoch AI · Aug 2025
4 more boards (other shared and single-model boards)
Gemma 3 12B IT
1341.7#84/188 open
ArenaLMArena · Oct 2026
GPT OSS 20B
1317.3#98/188 open
ArenaLMArena · Oct 2026
Only GPT OSS 20B (3)
- Terminal-Bench3.4%
- WeirdML40.9%
- ALE-Bench566.0
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 3 12B IT
5.7 GB4.8 GB file
32K 11.9 GB · 128K —
GPT OSS 20B
9.3 GB11.5 GB file
32K 9.8 GB · 128K 11.5 GB
Q3_K_S
Gemma 3 12B IT
5.9 GB5.5 GB file
32K 12.1 GB · 128K —
GPT OSS 20B
9.5 GB11.5 GB file
32K 10.0 GB · 128K 11.7 GB
Q3_K_M
Gemma 3 12B IT
6.5 GB6.0 GB file
32K 12.7 GB · 128K —
GPT OSS 20B
10.6 GB11.5 GB file
32K 11.1 GB · 128K 12.8 GB
Q4_0
Gemma 3 12B IT
6.7 GB6.9 GB file
32K 12.9 GB · 128K —
GPT OSS 20B
10.8 GB11.5 GB file
32K 11.3 GB · 128K 13.0 GB
Q4_K_M
Gemma 3 12B IT
8.0 GB7.3 GB file
32K 14.2 GB · 128K —
GPT OSS 20B
12.9 GB11.6 GB file
32K 13.4 GB · 128K 15.1 GB
Q5_K_M
Gemma 3 12B IT
9.6 GB8.4 GB file
32K 15.7 GB · 128K —
GPT OSS 20B
15.3 GB11.7 GB file
32K 15.8 GB · 128K 17.5 GB
Q6_K
Gemma 3 12B IT
11.1 GB9.7 GB file
32K 17.3 GB · 128K —
GPT OSS 20B
17.6 GB12.0 GB file
32K 18.1 GB · 128K 19.8 GB
Q8_0
Gemma 3 12B IT
13.4 GB12.5 GB file
32K 19.6 GB · 128K —
GPT OSS 20B
21.3 GB~20.9 GB file
32K 21.8 GB · 128K 23.5 GB
19 more quants
IQ2_XXS
Gemma 3 12B IT
3.7 GB3.6 GB file
32K 9.9 GB · 128K —
GPT OSS 20B
6.1 GB~5.8 GB file
32K 6.6 GB · 128K 8.3 GB
IQ2_XS
Gemma 3 12B IT
4.0 GB3.8 GB file
32K 10.2 GB · 128K —
GPT OSS 20B
6.7 GB~6.3 GB file
32K 7.2 GB · 128K 8.9 GB
IQ2_S
Gemma 3 12B IT
4.2 GB4.0 GB file
32K 10.4 GB · 128K —
GPT OSS 20B
6.9 GB~6.5 GB file
32K 7.4 GB · 128K 9.1 GB
IQ2_M
Gemma 3 12B IT
4.5 GB4.4 GB file
32K 10.7 GB · 128K —
GPT OSS 20B
7.4 GB~7.1 GB file
32K 7.9 GB · 128K 9.6 GB
IQ3_XXS
Gemma 3 12B IT
5.2 GB4.8 GB file
32K 11.4 GB · 128K —
GPT OSS 20B
8.5 GB11.6 GB file
32K 9.0 GB · 128K 10.7 GB
Q2_K_S
Gemma 3 12B IT
5.4 GB4.4 GB file
32K 11.6 GB · 128K —
GPT OSS 20B
8.7 GB12.1 GB file
32K 9.2 GB · 128K 10.9 GB
IQ3_XS
Gemma 3 12B IT
5.5 GB5.2 GB file
32K 11.7 GB · 128K —
GPT OSS 20B
9.0 GB11.6 GB file
32K 9.5 GB · 128K 11.2 GB
IQ3_S
Gemma 3 12B IT
5.7 GB5.2 GB file
32K 11.9 GB · 128K —
GPT OSS 20B
9.3 GB12.1 GB file
32K 9.8 GB · 128K 11.5 GB
IQ3_M
Gemma 3 12B IT
6.0 GB5.7 GB file
32K 12.2 GB · 128K —
GPT OSS 20B
9.8 GB11.6 GB file
32K 10.3 GB · 128K 12.0 GB
Q3_K_L
Gemma 3 12B IT
6.9 GB6.5 GB file
32K 13.1 GB · 128K —
GPT OSS 20B
11.1 GB11.5 GB file
32K 11.6 GB · 128K 13.3 GB
IQ4_XS
Gemma 3 12B IT
7.2 GB6.6 GB file
32K 13.4 GB · 128K —
GPT OSS 20B
11.6 GB11.6 GB file
32K 12.1 GB · 128K 13.8 GB
IQ4_NL
Gemma 3 12B IT
7.5 GB6.9 GB file
32K 13.7 GB · 128K —
GPT OSS 20B
12.1 GB11.6 GB file
32K 12.6 GB · 128K 14.3 GB
Q4_1
Gemma 3 12B IT
7.5 GB7.6 GB file
32K 13.7 GB · 128K —
GPT OSS 20B
12.1 GB11.6 GB file
32K 12.6 GB · 128K 14.3 GB
Q4_K_S
Gemma 3 12B IT
7.5 GB6.9 GB file
32K 13.7 GB · 128K —
GPT OSS 20B
12.1 GB11.6 GB file
32K 12.6 GB · 128K 14.3 GB
Q4_K_L
Gemma 3 12B IT
8.2 GB7.5 GB file
32K 14.4 GB · 128K —
GPT OSS 20B
13.2 GB11.9 GB file
32K 13.7 GB · 128K 15.4 GB
Q5_0
Gemma 3 12B IT
8.4 GB8.2 GB file
32K 14.6 GB · 128K —
GPT OSS 20B
not listed
Q5_K_S
Gemma 3 12B IT
9.2 GB8.2 GB file
32K 15.4 GB · 128K —
GPT OSS 20B
14.8 GB11.7 GB file
32K 15.3 GB · 128K 17.0 GB
Q5_K_L
Gemma 3 12B IT
9.7 GB8.7 GB file
32K 15.9 GB · 128K —
GPT OSS 20B
15.5 GB11.9 GB file
32K 16.0 GB · 128K 17.7 GB
BF16
Gemma 3 12B IT
26.8 GB23.5 GB file
32K 33.0 GB · 128K —
GPT OSS 20B
42.2 GB~41.8 GB file
32K 42.7 GB · 128K 44.4 GB
Gemma 3 12B IT
Ollama · Q4_K_M
ollama run hf.co/unsloth/gemma-3-12b-it-GGUF:Q4_K_MGPT OSS 20B
Ollama · Q4_K_M
ollama run hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M| Spec | Gemma 3 12B IT | GPT OSS 20B |
|---|---|---|
| Publisher | OpenAI | |
| Parameters | 12.2B | 20.9B |
| Active parameters | — | 4.2B per token |
| Architecture | — | GptOssForCausalLM |
| Context | 32K tokens | 128K tokens |
| Modalities | Vision | Chat, Function-calling |
| License | Gemma Terms | Apache 2.0 |
| Released | Mar 2025 | Aug 2025 |
Frequently Asked Questions
Which is better, Gemma 3 12B IT or GPT OSS 20B?
GPT OSS 20B leads llmrun Score 96.4 to 74.2; both fit 24 GB at Q4_K_M; GPT OSS 20B is 2.3× faster on RTX 4090 (~190 vs ~82 tok/s). Choose Gemma 3 12B IT for lighter on memory and also does vision; choose GPT OSS 20B for higher llmrun Score and stronger at reasoning.
Which needs less VRAM, Gemma 3 12B IT or GPT OSS 20B?
At Q4_K_M, Gemma 3 12B IT requires 8.0 GB while GPT OSS 20B needs 12.9 GB, making Gemma 3 12B IT the easier model to run on your local setup.
Which is faster on RTX 4090?
Estimated decode speed on RTX 4090 is about 82 tok/s for Gemma 3 12B IT (Q4_K_M) and 190 tok/s for GPT OSS 20B (Q4_K_M), so GPT OSS 20B is faster. Estimates, ±20%.
Can I run Gemma 3 12B IT or GPT OSS 20B on a 16 GB GPU?
Gemma 3 12B IT fits at Q4_K_M; GPT OSS 20B fits at Q4_K_M.
Which has a longer context window, Gemma 3 12B IT or GPT OSS 20B?
Gemma 3 12B IT supports 32,768 tokens and GPT OSS 20B supports 131,072 tokens.