In December 2025 Nous Research stated that Hermes 4.3 (36B) is its first production model post-trained entirely on Psyche, that the run averaged 144k tokens per second across 24 Psyche nodes, and that the Psyche-trained version outperformed a centrally trained version on a suite of downstream tasks.
Supported
Project announcement
Evidence (3): 3 supporting
- supports: Nous Research
Hermes 4.3 is our first production model post-trained entirely on the Psyche network
- supports: Nous Research
The training run proved stable throughout, averaging 144k tokens/second spread across 24 Psyche nodes.
- supports: Nous Research
The Psyche trained version of Hermes 4.3 outperformed the traditional centralized version on a suite of downstream tasks.
Whether Psyche's smart contracts are deployed on Solana mainnet is not yet verifiable. Nous's posts refer to a Psyche testnet (September 2025) and link a devnet address (December 2025), and no source checked states a mainnet deployment.
Not yet verifiable
Evidence (2): 2 context
- context: Nous Research
https://explorer.solana.com/address/HR8RN2TP9E9zsi2kjhvPbirJWA1R6L6ruf4xNNGpjU5Y?cluster=devnet
- context: Nous Research
Since the launch of the Psyche testnet we've made substantial improvements to the codebase
In a post dated 29 September 2025 Nous Research stated that its initial Psyche training run on testnet, the Consilience 40B run, was the largest distributed pre-training run ever.
Supported
Project announcement
Evidence (1): 1 supporting
- supports: Nous Research
Our initial training run on testnet verified that we can train models over internet bandwidth at significant parameter and dataset sizes. By both measures, our Consilience 40B run was the largest distributed pre-training run ever.