llmrun

GPT OSS 20B vs Qwen3 8B

GPT OSS 20B leads llmrun Score 96.2 to 93.8 (ranges overlap); both fit 24 GB at Q4_K_M; GPT OSS 20B is 1.6× faster on RTX 4090 (~190 vs ~119 tok/s).

Model A · OpenAI

GPT OSS 20B
96.2llmrun Score

20.9B · 4.2B active · 128K context · Apache 2.0 · Aug 2025

ChatFunctions

Model B · Alibaba

Qwen3 8B
93.8llmrun Score

8.2B · 40K context · Apache 2.0 · Apr 2025

ChatFunctions

Which should you choose?

Choose GPT OSS 20B if you want…

  • Higher llmrun Score

    96.2 vs 93.8, ranges overlap

  • Stronger at reasoning

    90.8 vs 81.1

  • Ahead on the benchmarks

    6 of 7 shared boards

  • 1.6× faster on RTX 4090

    ~190 vs ~119 tok/s

  • Longer context window

    128K vs 40K tokens

  • Newer release

    Aug 2025 vs Apr 2025

Choose Qwen3 8B if you want…

  • Fits an 8 GB GPU

    Q4_K_M · 7.6 GB; GPT OSS 20B needs 6.3 GB at its smallest quant

  • Lighter on memory

    5.5 vs 12.9 GB at Q4_K_M

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

GPT OSS 20B

IQ2_M · ~103 tok/s

7.6 GB · up to 32K

Qwen3 8B

Q4_K_M · ~32 tok/s

7.6 GB · up to 16Kmore headroom

12 GB

RTX 3060 12GB · 12 GB tier

GPT OSS 20B

IQ4_XS · ~89 tok/s

11.8 GB · up to 16K

Qwen3 8B

Q4_K_M · ~42 tok/s

7.6 GB · up to 40Kmore headroom

16 GB

RTX 5060 Ti 16GB · 16 GB tier

GPT OSS 20B

Q4_K_M · ~98 tok/s

13.1 GB · up to 128Kfaster

Qwen3 8B

Q4_K_M · ~53 tok/s

7.6 GB · up to 40K

24 GBreference

RTX 4090 · 24 GB tier

GPT OSS 20B

Q4_K_M · ~190 tok/s

13.1 GB · up to 128Kfaster

Qwen3 8B

Q4_K_M · ~119 tok/s

7.6 GB · up to 40K

32 GB

RTX 5090 · 32 GB tier

GPT OSS 20B

Q4_K_M · ~282 tok/s

13.1 GB · up to 128Kfaster

Qwen3 8B

Q4_K_M · ~211 tok/s

7.6 GB · up to 40K

48 GB

RTX A6000 · 48 GB tier

GPT OSS 20B

Q4_K_M · ~154 tok/s

13.1 GB · up to 128Kfaster

Qwen3 8B

Q4_K_M · ~90 tok/s

7.6 GB · up to 40K

Mac 32 GB

Mac Mini M4 · Macs

GPT OSS 20B

Q4_K_M · ~28 tok/s

13.1 GB · up to 128Kfaster

Qwen3 8B

Q4_K_M · ~15 tok/s

7.6 GB · up to 40K

Mac 64 GB

Mac Studio M4 Max · Macs

GPT OSS 20B

Q4_K_M · ~86 tok/s

13.1 GB · up to 128Kfaster

Qwen3 8B

Q4_K_M · ~69 tok/s

7.6 GB · up to 40K

Mac 128 GB

Mac Studio M4 Max · Macs

GPT OSS 20B

Q4_K_M · ~86 tok/s

13.1 GB · up to 128Kfaster

Qwen3 8B

Q4_K_M · ~69 tok/s

7.6 GB · up to 40K

Reference GPU: RTX 4090. Tinted cell = the better pick on that machine (fits, more headroom, or at least 10% faster).

llmrun Score

with its 90% range · bars share one scale

Overall

GPT OSS 20B

96.293–99

Qwen3 8B

93.891–97

Coding

GPT OSS 20B

100.992–110

Qwen3 8B

no data

Science

GPT OSS 20B

92.485–100

Qwen3 8B

87.780–95

Reasoning

GPT OSS 20B

90.882–99

Qwen3 8B

81.172–91

GPT OSS 20B · #58 of 92 open models · #15 of 34 that fit 24 GB

Qwen3 8B · #62 of 92 open models · #18 of 34 that fit 24 GB

Benchmarks

7 shared boards: GPT OSS 20B leads 6, Qwen3 8B leads 1 · Benchmarks imported Oct 2026

GPQA Diamond

GPT OSS 20B

60.8%#41/97 open

IndependentEpoch AI · Aug 2025

Qwen3 8B

56.8%#43/97 open

IndependentEpoch AI · Apr 2025

AIME 2024/2025

GPT OSS 20B

65.3%#33/85 open

IndependentEpoch AI · Aug 2025

Qwen3 8B

56.1%#39/85 open

IndependentEpoch AI · Apr 2025

SciCode

GPT OSS 20B

34.4%#47/65 open

Lab-reportedEpoch AI · Aug 2025

Qwen3 8B

22.6%#60/65 open

Lab-reportedEpoch AI · Apr 2025

Chess Puzzles

GPT OSS 20B

4.0%#33/67 open

IndependentEpoch AI · Aug 2025

Qwen3 8B

5.0%#31/67 open

IndependentEpoch AI · Apr 2025

CritPt

GPT OSS 20B

1.4%#30/65 open

Lab-reportedEpoch AI · Aug 2025

Qwen3 8B

0.0%#63/65 open

Lab-reportedEpoch AI · Apr 2025

DTBench

GPT OSS 20B

68.0%#34/68 open

Lab-reportedEpoch AI · Aug 2025

Qwen3 8B

59.7%#50/68 open

Lab-reportedEpoch AI · Apr 2025

LMCA

GPT OSS 20B

14.5%#43/54 open

Lab-reportedEpoch AI · Aug 2025

Qwen3 8B

8.8%#49/54 open

Lab-reportedEpoch AI · Apr 2025

6 more boards (other shared and single-model boards)

Only GPT OSS 20B (5)

Only Qwen3 8B (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

GPT OSS 20B

9.3 GB11.5 GB file

32K 9.8 GB · 128K 11.5 GB

Qwen3 8B

4.1 GB3.3 GB file

32K 8.6 GB · 128K —

Q3_K_S

GPT OSS 20B

9.5 GB11.5 GB file

32K 10.0 GB · 128K 11.7 GB

Qwen3 8B

4.2 GB3.8 GB file

32K 8.7 GB · 128K —

Q3_K_M

GPT OSS 20B

10.6 GB11.5 GB file

32K 11.1 GB · 128K 12.8 GB

Qwen3 8B

4.6 GB4.1 GB file

32K 9.1 GB · 128K —

Q4_0

GPT OSS 20B

10.8 GB11.5 GB file

32K 11.3 GB · 128K 13.0 GB

Qwen3 8B

4.7 GB4.8 GB file

32K 9.2 GB · 128K —

Q4_K_M

GPT OSS 20B

12.9 GB11.6 GB file

32K 13.4 GB · 128K 15.1 GB

Qwen3 8B

5.5 GB5.0 GB file

32K 10.1 GB · 128K —

Q5_K_M

GPT OSS 20B

15.3 GB11.7 GB file

32K 15.8 GB · 128K 17.5 GB

Qwen3 8B

6.4 GB5.9 GB file

32K 11.0 GB · 128K —

Q6_K

GPT OSS 20B

17.6 GB12.0 GB file

32K 18.1 GB · 128K 19.8 GB

Qwen3 8B

7.4 GB6.7 GB file

32K 11.9 GB · 128K —

Q8_0

GPT OSS 20B

21.3 GB~20.9 GB file

32K 21.8 GB · 128K 23.5 GB

Qwen3 8B

8.8 GB8.7 GB file

32K 13.3 GB · 128K —

19 more quants

IQ2_XXS

GPT OSS 20B

6.1 GB~5.8 GB file

32K 6.6 GB · 128K 8.3 GB

Qwen3 8B

2.9 GB2.6 GB file

32K 7.4 GB · 128K —

IQ2_XS

GPT OSS 20B

6.7 GB~6.3 GB file

32K 7.2 GB · 128K 8.9 GB

Qwen3 8B

3.1 GB2.7 GB file

32K 7.6 GB · 128K —

IQ2_S

GPT OSS 20B

6.9 GB~6.5 GB file

32K 7.4 GB · 128K 9.1 GB

Qwen3 8B

3.2 GB2.9 GB file

32K 7.7 GB · 128K —

IQ2_M

GPT OSS 20B

7.4 GB~7.1 GB file

32K 7.9 GB · 128K 9.6 GB

Qwen3 8B

3.4 GB3.1 GB file

32K 7.9 GB · 128K —

IQ3_XXS

GPT OSS 20B

8.5 GB11.6 GB file

32K 9.0 GB · 128K 10.7 GB

Qwen3 8B

3.8 GB3.4 GB file

32K 8.3 GB · 128K —

Q2_K_S

GPT OSS 20B

8.7 GB12.1 GB file

32K 9.2 GB · 128K 10.9 GB

Qwen3 8B

3.9 GB3.1 GB file

32K 8.4 GB · 128K —

IQ3_XS

GPT OSS 20B

9.0 GB11.6 GB file

32K 9.5 GB · 128K 11.2 GB

Qwen3 8B

4.0 GB3.6 GB file

32K 8.5 GB · 128K —

IQ3_S

GPT OSS 20B

9.3 GB12.1 GB file

32K 9.8 GB · 128K 11.5 GB

Qwen3 8B

4.1 GB3.8 GB file

32K 8.6 GB · 128K —

IQ3_M

GPT OSS 20B

9.8 GB11.6 GB file

32K 10.3 GB · 128K 12.0 GB

Qwen3 8B

4.3 GB3.9 GB file

32K 8.8 GB · 128K —

Q3_K_L

GPT OSS 20B

11.1 GB11.5 GB file

32K 11.6 GB · 128K 13.3 GB

Qwen3 8B

4.8 GB4.4 GB file

32K 9.3 GB · 128K —

IQ4_XS

GPT OSS 20B

11.6 GB11.6 GB file

32K 12.1 GB · 128K 13.8 GB

Qwen3 8B

5.0 GB4.6 GB file

32K 9.5 GB · 128K —

IQ4_NL

GPT OSS 20B

12.1 GB11.6 GB file

32K 12.6 GB · 128K 14.3 GB

Qwen3 8B

5.2 GB4.8 GB file

32K 9.7 GB · 128K —

Q4_1

GPT OSS 20B

12.1 GB11.6 GB file

32K 12.6 GB · 128K 14.3 GB

Qwen3 8B

5.2 GB5.2 GB file

32K 9.7 GB · 128K —

Q4_K_S

GPT OSS 20B

12.1 GB11.6 GB file

32K 12.6 GB · 128K 14.3 GB

Qwen3 8B

5.2 GB4.8 GB file

32K 9.7 GB · 128K —

Q4_K_L

GPT OSS 20B

13.2 GB11.9 GB file

32K 13.7 GB · 128K 15.4 GB

Qwen3 8B

5.6 GB5.5 GB file

32K 10.2 GB · 128K —

Q5_0

GPT OSS 20B

not listed

Qwen3 8B

5.7 GB5.7 GB file

32K 10.3 GB · 128K —

Q5_K_S

GPT OSS 20B

14.8 GB11.7 GB file

32K 15.3 GB · 128K 17.0 GB

Qwen3 8B

6.2 GB5.7 GB file

32K 10.8 GB · 128K —

Q5_K_L

GPT OSS 20B

15.5 GB11.9 GB file

32K 16.0 GB · 128K 17.7 GB

Qwen3 8B

6.5 GB6.2 GB file

32K 11.1 GB · 128K —

BF16

GPT OSS 20B

42.2 GB~41.8 GB file

32K 42.7 GB · 128K 44.4 GB

Qwen3 8B

17.0 GB16.4 GB file

32K 21.5 GB · 128K —

How to run each

Specifications

Specifications of GPT OSS 20B and Qwen3 8B
SpecGPT OSS 20BQwen3 8B
PublisherOpenAIAlibaba
Parameters20.9B8.2B
Active parameters4.2B per token—
ArchitectureGptOssForCausalLMQwen3ForCausalLM
Context128K tokens40K tokens
ModalitiesChat, Function-callingChat, Function-calling
LicenseApache 2.0Apache 2.0
ReleasedAug 2025Apr 2025

Frequently Asked Questions

Which is better, GPT OSS 20B or Qwen3 8B?

GPT OSS 20B leads llmrun Score 96.2 to 93.8 (ranges overlap); both fit 24 GB at Q4_K_M; GPT OSS 20B is 1.6× faster on RTX 4090 (~190 vs ~119 tok/s). Choose GPT OSS 20B for higher llmrun Score and stronger at reasoning; choose Qwen3 8B for fits an 8 GB GPU and lighter on memory.

Which needs less VRAM, GPT OSS 20B or Qwen3 8B?

At Q4_K_M, GPT OSS 20B requires 12.9 GB while Qwen3 8B needs 5.5 GB, making Qwen3 8B the easier model to run on your local setup.

Which is faster on RTX 4090?

Estimated decode speed on RTX 4090 is about 190 tok/s for GPT OSS 20B (Q4_K_M) and 119 tok/s for Qwen3 8B (Q4_K_M), so GPT OSS 20B is faster. Estimates, ±20%.

Can I run GPT OSS 20B or Qwen3 8B on a 16 GB GPU?

GPT OSS 20B fits at Q4_K_M; Qwen3 8B fits at Q4_K_M.

Which has a longer context window, GPT OSS 20B or Qwen3 8B?

GPT OSS 20B supports 131,072 tokens and Qwen3 8B supports 40,960 tokens.

GPT OSS 20B vs Qwen3 8B — VRAM & GPU Comparison | llmrun