A new economic primitive for funding decentralized AI

We track, rank and pay for the best open source decentralized LLMs to compete against OpenAI

1,873

LLM models

$326,734

Paid to data scientists

6,557

Developers

Projects integrated into the Arrakis AI Ecosystem

Crowdsourcing our collective intelligence to build the best AI

Open source AI has been lagging behind the likes of Google and OpenAI by billions of dollars.

Salt aims to solve that by rewarding open source developers who contribute to the democratization of AI. We run competitions between AI models to find and reward the best AI models. As a result, our users will be able to access the latest cutting edge AI models.

LLM Leaderboard

We evaluate LLMs on key benchmarks using the Eleuther AI, a framework to test LLMs on a large number of different evaluation tasks. The higher the score, the better the LLM.

Model Name

Average

ARC

HellaSwag

MMLU

TruthfulQA

Winogrande

GSM8K

Usage

meta-llama/Meta-Llama-3-70B-Instruct

77.88

71.42

85.69

80.06

61.81

82.87

85.44

1,190,580

saltlux/luxia-21.4b-alignment-v1.0

77.74

77.47

91.88

68.1

79.17

87.37

71.11

1,185,234

zhengr/MixTAO-7Bx2-MoE-v8.1

77.5

73.81

89.22

64.92

78.57

87.37

71.11

1,178,889

yungconglong/Truthful_DPO_TomGrc_FusionNet

77.44

74.91

89.3

64.67

78.02

88.24

69.52

1,177,065

JaeyeonKang/CCK_Asura_v1

77.43

73.89

89.07

75.44

71.75

86.35

68.08

1,170,954

fblgit/UNA-SimpleSmaug-34b-v1beta

77.41

74.57

86.74

76.68

70.17

83.82

72.48

1,162,959

TomGrc/FusionNet_34Bx2_MoE_v0.1

77.38

73.72

86.46

76.72

71.01

83.35

73.01

1,144,223

migitssera/Tess-72B-v1.5b

77.3

71.25

85.53

81.22

71.99

81.45

76.95

1,129,110

Join our community

Join us on our mission to to the moon & revolutionize open source AI development so that we can build a permissionless, democratized, and decentralized AI.

Let the fate of AI be in our hands and not that of big tech companies.

Join our community and harvest $SALT