Modelle & Konfigurator

Welches Modell passt zu dir?

Alle Modelle die wir lokal deployen. Wähle ein Modell, konfiguriere deine Hardware und berechne die Kosten.

DeepSeek V3.2

moe

DeepSeek AI · 685B (37B aktiv)

Layers

61

Context

128K

Q4

404.0GB

Q8

715.0GB

GLM-5

moe

Z.ai (Zhipu AI) · 744B (40B aktiv)

Layers

78

Context

198K

Q4

457.0GB

Q8

801.0GB

gpt-oss-120b

moe

OpenAI · 117B (5.1B aktiv)

Layers

36

Context

128K

Q4

63.0GB

Q8

63.0GB

gpt-oss-20b

moe

OpenAI · 21B (3.6B aktiv)

Layers

24

Context

128K

Q4

12.0GB

Q8

12.0GB

Kimi K2.5

moe

Moonshot AI · 1T (32B aktiv)

Layers

61

Context

256K

Q4

621.0GB

Q8

1090.0GB

MiMo-V2-Flash

moe

Xiaomi · 309B (15B aktiv)

Layers

48

Context

256K

Q4

187.0GB

Q8

328.0GB

MiniMax-M2.5

moe

MiniMax · 230B (10B aktiv)

Layers

62

Context

200K

Q4

140.0GB

Q8

243.0GB

Nemotron 3 Nano 30B-A3B

moe

NVIDIA · 30B (3.5B aktiv)

Layers

52

Context

1024K

Q4

24.0GB

Q8

36.0GB

Nemotron 3 Super 120B-A12B

moe

NVIDIA · 120B (12B aktiv)

Layers

88

Context

1024K

Q4

66.0GB

Q8

67.0GB

Qwen3.5-122B-A10B

moe

Alibaba / Qwen · 122B (10B aktiv)

Layers

48

Context

256K

Q4

75.0GB

Q8

130.0GB

Qwen3.5-27B

dense

Alibaba / Qwen · 27B (27B aktiv)

Layers

64

Context

256K

Q4

17.0GB

Q8

30.0GB

Qwen3.5-35B-A3B

moe

Alibaba / Qwen · 35B (3B aktiv)

Layers

40

Context

256K

Q4

22.0GB

Q8

37.0GB

Qwen3.5-397B-A17B

moe

Alibaba / Qwen · 397B (17B aktiv)

Layers

60

Context

256K

Q4

242.0GB

Q8

428.0GB

Step-3.5-Flash

moe

StepFun · 196B (11B aktiv)

Layers

45

Context

256K

Q4

120.0GB

Q8

209.0GB

Benchmarks

Modellvergleich

Alle Werte von offiziellen Model Cards und Technical Reports.

Overall (Durchschnitt)

Qwen3.5-397B-A17B

72.3

Kimi K2.5

70.9

Qwen3.5-122B-A10B

69.7

DeepSeek V3.2

68.9

Qwen3.5-27B

68.4

Step-3.5-Flash

68.0

Qwen3.5-35B-A3B

65.9

MiniMax-M2.5

63.8

Nemotron 3 Super 120B-A12B

62.5

MiMo-V2-Flash

62.2

gpt-oss-120b

60.4

GLM-5

59.0

gpt-oss-20b

53.5

Nemotron 3 Nano 30B-A3B

53.4

GPQA-Diamond

Qwen3.5-397B-A17B

88.4

Kimi K2.5

87.6

Qwen3.5-122B-A10B

86.6

DeepSeek V3.2

82.4

Qwen3.5-27B

85.5

Step-3.5-Flash

83.1

Qwen3.5-35B-A3B

84.2

MiniMax-M2.5

85.2

Nemotron 3 Super 120B-A12B

79.2

MiMo-V2-Flash

83.7

gpt-oss-120b

80.1

GLM-5

86.0

gpt-oss-20b

71.5

Nemotron 3 Nano 30B-A3B

73.0

SWE-Bench

Qwen3.5-397B-A17B

76.4

Kimi K2.5

76.8

Qwen3.5-122B-A10B

72.0

DeepSeek V3.2

73.1

Qwen3.5-27B

72.4

Step-3.5-Flash

74.4

Qwen3.5-35B-A3B

69.2

MiniMax-M2.5

80.2

Nemotron 3 Super 120B-A12B

60.5

MiMo-V2-Flash

73.4

gpt-oss-120b

62.0

GLM-5

77.8

gpt-oss-20b

54.2

Nemotron 3 Nano 30B-A3B

38.8

HumanEval

Qwen3.5-397B-A17B

97.1

Kimi K2.5

99.0

Qwen3.5-122B-A10B

96.3

DeepSeek V3.2

93.0

Qwen3.5-27B

96.0

Step-3.5-Flash

81.1

Qwen3.5-35B-A3B

95.2

MiniMax-M2.5

92.0

Nemotron 3 Super 120B-A12B

88.3

MiMo-V2-Flash

80.6

gpt-oss-120b

90.2

GLM-5

97.0

gpt-oss-20b

93.1

Nemotron 3 Nano 30B-A3B

82.4

MATH

Qwen3.5-397B-A17B

97.8

Kimi K2.5

82.0

Qwen3.5-122B-A10B

96.5

DeepSeek V3.2

97.0

Qwen3.5-27B

96.2

Step-3.5-Flash

97.0

Qwen3.5-35B-A3B

95.8

MiniMax-M2.5

78.0

Nemotron 3 Super 120B-A12B

96.5

MiMo-V2-Flash

92.0

gpt-oss-120b

97.6

GLM-5

97.4

gpt-oss-20b

97.8

Nemotron 3 Nano 30B-A3B

95.2

MMMU-Pro

Qwen3.5-397B-A17B

79.0

Kimi K2.5

78.5

Qwen3.5-122B-A10B

76.9

DeepSeek V3.2

85.0

Qwen3.5-27B

75.0

Step-3.5-Flash

83.3

Qwen3.5-35B-A3B

75.1

MiniMax-M2.5

52.0

Nemotron 3 Super 120B-A12B

62.8

MiMo-V2-Flash

84.9

gpt-oss-120b

58.7

GLM-5

0.0

gpt-oss-20b

47.6

Nemotron 3 Nano 30B-A3B

52.1

LiveCodeBench

Qwen3.5-397B-A17B

83.6

Kimi K2.5

85.0

Qwen3.5-122B-A10B

78.9

DeepSeek V3.2

83.3

Qwen3.5-27B

80.7

Step-3.5-Flash

86.4

Qwen3.5-35B-A3B

74.6

MiniMax-M2.5

65.0

Nemotron 3 Super 120B-A12B

78.7

MiMo-V2-Flash

80.6

gpt-oss-120b

82.7

GLM-5

52.0

gpt-oss-20b

68.3

Nemotron 3 Nano 30B-A3B

68.3

IFBench

Qwen3.5-397B-A17B

76.5

Kimi K2.5

70.2

Qwen3.5-122B-A10B

76.1

DeepSeek V3.2

65.0

Qwen3.5-27B

76.5

Step-3.5-Flash

64.6

Qwen3.5-35B-A3B

70.2

MiniMax-M2.5

70.0

Nemotron 3 Super 120B-A12B

72.6

MiMo-V2-Flash

39.9

gpt-oss-120b

69.0

GLM-5

46.5

gpt-oss-20b

61.4

Nemotron 3 Nano 30B-A3B

71.5

HLE

Qwen3.5-397B-A17B

28.7

Kimi K2.5

30.1

Qwen3.5-122B-A10B

25.3

DeepSeek V3.2

25.1

Qwen3.5-27B

24.3

Step-3.5-Flash

19.1

Qwen3.5-35B-A3B

22.4

MiniMax-M2.5

19.4

Nemotron 3 Super 120B-A12B

18.3

MiMo-V2-Flash

22.1

gpt-oss-120b

14.9

GLM-5

30.5

gpt-oss-20b

9.2

Nemotron 3 Nano 30B-A3B

10.6

SciCode

Qwen3.5-397B-A17B

43.2

Kimi K2.5

48.7

Qwen3.5-122B-A10B

38.7

DeepSeek V3.2

38.9

Qwen3.5-27B

35.4

Step-3.5-Flash

40.4

Qwen3.5-35B-A3B

32.1

MiniMax-M2.5

44.4

Nemotron 3 Super 120B-A12B

42.0

MiMo-V2-Flash

25.9

gpt-oss-120b

29.8

GLM-5

46.2

gpt-oss-20b

22.4

Nemotron 3 Nano 30B-A3B

33.3

Terminal-Bench

Qwen3.5-397B-A17B

52.5

Kimi K2.5

50.8

Qwen3.5-122B-A10B

49.4

DeepSeek V3.2

46.4

Qwen3.5-27B

41.6

Step-3.5-Flash

51.0

Qwen3.5-35B-A3B

40.5

MiniMax-M2.5

51.7

Nemotron 3 Super 120B-A12B

25.8

MiMo-V2-Flash

38.5

gpt-oss-120b

18.7

GLM-5

56.2

gpt-oss-20b

9.8

Nemotron 3 Nano 30B-A3B

8.5

Dein Modell ist nicht dabei? Kontaktiere uns. Wir deployen jedes Open-Source Modell.