Magistral Small 2509 vs GPT OSS 20B
GPT OSS 20B leads llmrun Score 96.4 to 85.6; both fit 24 GB at Q4_K_M; GPT OSS 20B is 4.4× faster on RTX 4090 (~190 vs ~43 tok/s).
Model A · Mistral AI
Magistral Small 250924.0B · 128K context · Apache 2.0 · Sep 2025
Model B · OpenAI
GPT OSS 20B20.9B · 4.2B active · 128K context · Apache 2.0 · Aug 2025
Choose GPT OSS 20B if you want…
Higher llmrun Score
96.4 vs 85.6
Stronger at science
93.1 vs 82.8
Ahead on the benchmarks
5 of 5 shared boards
Fits a 12 GB GPU
IQ4_XS · 11.8 GB; Magistral Small 2509 needs 10.3 GB at its smallest quant
4.4× faster on RTX 4090
~190 vs ~43 tok/s
Also does function-calling
Magistral Small 2509 doesn't list it
Magistral Small 2509: no clear edge over GPT OSS 20B in the data we have.
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
Magistral Small 2509
Doesn't fit
smallest quant needs 10.3 GB
GPT OSS 20B
IQ2_M · ~103 tok/s
7.6 GB · up to 32Kfits
12 GB
RTX 3060 12GB · 12 GB tier
Magistral Small 2509
IQ2_M · ~27 tok/s
11.8 GB · up to 16K
GPT OSS 20B
IQ4_XS · ~89 tok/s
11.8 GB · up to 16Kbetter quant
16 GB
RTX 5060 Ti 16GB · 16 GB tier
Magistral Small 2509
Q4_K_M · ~19 tok/stight
18.1 GB · up to 4K
GPT OSS 20B
Q4_K_M · ~98 tok/s
13.1 GB · up to 128Kmore headroom
24 GBreference
RTX 4090 · 24 GB tier
Magistral Small 2509
Q4_K_M · ~43 tok/s
18.1 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
Magistral Small 2509
Q4_K_M · ~77 tok/s
18.1 GB · up to 64K
GPT OSS 20B
Q4_K_M · ~282 tok/s
13.1 GB · up to 128Kfaster
48 GB
RTX A6000 · 48 GB tier
Magistral Small 2509
Q4_K_M · ~33 tok/s
18.1 GB · up to 128K
GPT OSS 20B
Q4_K_M · ~154 tok/s
13.1 GB · up to 128Kfaster
Mac 32 GB
Mac Mini M4 · Macs
Magistral Small 2509
Q4_K_M · ~6 tok/s
18.1 GB · up to 64K
GPT OSS 20B
Q4_K_M · ~28 tok/s
13.1 GB · up to 128Kfaster
Mac 64 GB
Mac Studio M4 Max · Macs
Magistral Small 2509
Q4_K_M · ~25 tok/s
18.1 GB · up to 128K
GPT OSS 20B
Q4_K_M · ~86 tok/s
13.1 GB · up to 128Kfaster
Mac 128 GB
Mac Studio M4 Max · Macs
Magistral Small 2509
Q4_K_M · ~25 tok/s
18.1 GB · up to 128K
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
Magistral Small 2509
GPT OSS 20B
Coding
Magistral Small 2509
no data
GPT OSS 20B
Science
Magistral Small 2509
GPT OSS 20B
Reasoning
Magistral Small 2509
no data
GPT OSS 20B
Magistral Small 2509 · #80 of 106 open models · #22 of 36 that fit 24 GB
GPT OSS 20B · #69 of 106 open models · #16 of 36 that fit 24 GB
Benchmarks
5 shared boards: Magistral Small 2509 leads 0, GPT OSS 20B leads 5 · Benchmarks imported Nov 2026
Magistral Small 2509
47.6%#60/100 open
IndependentEpoch AI · Sep 2025
GPT OSS 20B
60.8%#44/100 open
IndependentEpoch AI · Aug 2025
Magistral Small 2509
28.1%#53/88 open
IndependentEpoch AI · Sep 2025
GPT OSS 20B
65.3%#36/88 open
IndependentEpoch AI · Aug 2025
Magistral Small 2509
35.2%#55/72 open
Lab-reportedEpoch AI · Sep 2025
GPT OSS 20B
38.9%#41/72 open
Lab-reportedEpoch AI · Aug 2025
Magistral Small 2509
0.3%#48/84 open
Lab-reportedEpoch AI · Sep 2025
GPT OSS 20B
1.4%#36/84 open
Lab-reportedEpoch AI · Aug 2025
Magistral Small 2509
61.3%#46/70 open
Lab-reportedEpoch AI · Sep 2025
GPT OSS 20B
68.0%#36/70 open
Lab-reportedEpoch AI · Aug 2025
6 more boards (other shared and single-model boards)
Only GPT OSS 20B (6)
- LMArena Text1317.3
- Terminal-Bench3.4%
- MMLU-Pro73.6%
- WeirdML40.9%
- LMCA14.5%
- 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
Magistral Small 2509
10.9 GB8.9 GB file
32K 17.2 GB · 128K 37.3 GB
GPT OSS 20B
9.3 GB11.5 GB file
32K 9.8 GB · 128K 11.5 GB
Q3_K_S
Magistral Small 2509
11.2 GB10.4 GB file
32K 17.5 GB · 128K 37.6 GB
GPT OSS 20B
9.5 GB11.5 GB file
32K 10.0 GB · 128K 11.7 GB
Q3_K_M
Magistral Small 2509
12.4 GB11.5 GB file
32K 18.7 GB · 128K 38.8 GB
GPT OSS 20B
10.6 GB11.5 GB file
32K 11.1 GB · 128K 12.8 GB
Q4_0
Magistral Small 2509
12.7 GB13.5 GB file
32K 19.0 GB · 128K 39.2 GB
GPT OSS 20B
10.8 GB11.5 GB file
32K 11.3 GB · 128K 13.0 GB
Q4_K_M
Magistral Small 2509
15.1 GB14.3 GB file
32K 21.4 GB · 128K 41.6 GB
GPT OSS 20B
12.9 GB11.6 GB file
32K 13.4 GB · 128K 15.1 GB
Q5_K_M
Magistral Small 2509
17.8 GB16.8 GB file
32K 24.1 GB · 128K 44.3 GB
GPT OSS 20B
15.3 GB11.7 GB file
32K 15.8 GB · 128K 17.5 GB
Q6_K
Magistral Small 2509
20.5 GB19.3 GB file
32K 26.8 GB · 128K 47.0 GB
GPT OSS 20B
17.6 GB12.0 GB file
32K 18.1 GB · 128K 19.8 GB
Q8_0
Magistral Small 2509
24.7 GB25.1 GB file
32K 31.0 GB · 128K 51.2 GB
GPT OSS 20B
21.3 GB~20.9 GB file
32K 21.8 GB · 128K 23.5 GB
18 more quants
IQ2_XXS
Magistral Small 2509
7.3 GB6.8 GB file
32K 13.6 GB · 128K 33.7 GB
GPT OSS 20B
6.1 GB~5.8 GB file
32K 6.6 GB · 128K 8.3 GB
IQ2_XS
Magistral Small 2509
7.9 GB7.2 GB file
32K 14.2 GB · 128K 34.3 GB
GPT OSS 20B
6.7 GB~6.3 GB file
32K 7.2 GB · 128K 8.9 GB
IQ2_S
Magistral Small 2509
8.2 GB7.5 GB file
32K 14.5 GB · 128K 34.6 GB
GPT OSS 20B
6.9 GB~6.5 GB file
32K 7.4 GB · 128K 9.1 GB
IQ2_M
Magistral Small 2509
8.8 GB8.2 GB file
32K 15.1 GB · 128K 35.2 GB
GPT OSS 20B
7.4 GB~7.1 GB file
32K 7.9 GB · 128K 9.6 GB
IQ3_XXS
Magistral Small 2509
10.0 GB9.4 GB file
32K 16.3 GB · 128K 36.4 GB
GPT OSS 20B
8.5 GB11.6 GB file
32K 9.0 GB · 128K 10.7 GB
Q2_K_S
Magistral Small 2509
not listed
GPT OSS 20B
8.7 GB12.1 GB file
32K 9.2 GB · 128K 10.9 GB
IQ3_XS
Magistral Small 2509
10.6 GB9.9 GB file
32K 16.9 GB · 128K 37.0 GB
GPT OSS 20B
9.0 GB11.6 GB file
32K 9.5 GB · 128K 11.2 GB
IQ3_S
Magistral Small 2509
not listed
GPT OSS 20B
9.3 GB12.1 GB file
32K 9.8 GB · 128K 11.5 GB
IQ3_M
Magistral Small 2509
11.5 GB10.7 GB file
32K 17.8 GB · 128K 37.9 GB
GPT OSS 20B
9.8 GB11.6 GB file
32K 10.3 GB · 128K 12.0 GB
Q3_K_L
Magistral Small 2509
13.0 GB12.4 GB file
32K 19.3 GB · 128K 39.5 GB
GPT OSS 20B
11.1 GB11.5 GB file
32K 11.6 GB · 128K 13.3 GB
IQ4_XS
Magistral Small 2509
13.6 GB12.8 GB file
32K 19.9 GB · 128K 40.1 GB
GPT OSS 20B
11.6 GB11.6 GB file
32K 12.1 GB · 128K 13.8 GB
IQ4_NL
Magistral Small 2509
14.2 GB13.5 GB file
32K 20.5 GB · 128K 40.7 GB
GPT OSS 20B
12.1 GB11.6 GB file
32K 12.6 GB · 128K 14.3 GB
Q4_1
Magistral Small 2509
14.2 GB14.9 GB file
32K 20.5 GB · 128K 40.7 GB
GPT OSS 20B
12.1 GB11.6 GB file
32K 12.6 GB · 128K 14.3 GB
Q4_K_S
Magistral Small 2509
14.2 GB13.5 GB file
32K 20.5 GB · 128K 40.7 GB
GPT OSS 20B
12.1 GB11.6 GB file
32K 12.6 GB · 128K 14.3 GB
Q4_K_L
Magistral Small 2509
15.4 GB14.8 GB file
32K 21.7 GB · 128K 41.9 GB
GPT OSS 20B
13.2 GB11.9 GB file
32K 13.7 GB · 128K 15.4 GB
Q5_K_S
Magistral Small 2509
17.2 GB16.3 GB file
32K 23.5 GB · 128K 43.7 GB
GPT OSS 20B
14.8 GB11.7 GB file
32K 15.3 GB · 128K 17.0 GB
Q5_K_L
Magistral Small 2509
18.1 GB17.2 GB file
32K 24.4 GB · 128K 44.6 GB
GPT OSS 20B
15.5 GB11.9 GB file
32K 16.0 GB · 128K 17.7 GB
BF16
Magistral Small 2509
48.7 GB47.2 GB file
32K 55.0 GB · 128K 75.2 GB
GPT OSS 20B
42.2 GB~41.8 GB file
32K 42.7 GB · 128K 44.4 GB
Magistral Small 2509
Ollama · Q4_K_M
ollama run hf.co/unsloth/Magistral-Small-2509-GGUF:Q4_K_MGPT OSS 20B
Ollama · Q4_K_M
ollama run hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M| Spec | Magistral Small 2509 | GPT OSS 20B |
|---|---|---|
| Publisher | Mistral AI | OpenAI |
| Parameters | 24.0B | 20.9B |
| Active parameters | — | 4.2B per token |
| Architecture | Mistral3ForConditionalGeneration | GptOssForCausalLM |
| Context | 128K tokens | 128K tokens |
| Modalities | Chat | Chat, Function-calling |
| License | Apache 2.0 | Apache 2.0 |
| Released | Sep 2025 | Aug 2025 |
Frequently Asked Questions
Which is better, Magistral Small 2509 or GPT OSS 20B?
GPT OSS 20B leads llmrun Score 96.4 to 85.6; both fit 24 GB at Q4_K_M; GPT OSS 20B is 4.4× faster on RTX 4090 (~190 vs ~43 tok/s). Choose GPT OSS 20B for higher llmrun Score and stronger at science.
Which needs less VRAM, Magistral Small 2509 or GPT OSS 20B?
At Q4_K_M, Magistral Small 2509 requires 15.1 GB while GPT OSS 20B needs 12.9 GB, making GPT OSS 20B the easier model to run on your local setup.
Which is faster on RTX 4090?
Estimated decode speed on RTX 4090 is about 43 tok/s for Magistral Small 2509 (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 Magistral Small 2509 or GPT OSS 20B on a 16 GB GPU?
Magistral Small 2509 fits at Q4_K_M; GPT OSS 20B fits at Q4_K_M.
Which has a longer context window, Magistral Small 2509 or GPT OSS 20B?
Magistral Small 2509 supports 131,072 tokens and GPT OSS 20B supports 131,072 tokens.