Together AI Joins Pearl Labs to Cut AI Inference Costs With Blockchain
Timothy Morano May 15, 2026 17:07
Together AI partners with Pearl Research Labs to slash AI inference costs using Proof of Useful Work, generating crypto rewards for GPU workloads.
Together AI has teamed up with Pearl Research Labs to tackle one of the most pressing challenges in artificial intelligence: the high cost of inference. The partnership introduces a discounted inference endpoint for the Gemma-4-31B-it-pearl model, leveraging Pearl’s blockchain protocol to turn AI computations into cryptocurrency emissions. Together AI claims this setup reduces inference costs by over 25%, a significant potential savings for enterprises deploying large-scale AI models.
At the core of Pearl’s approach is Proof of Useful Work (PoUW), a blockchain consensus mechanism designed to replace traditional energy-wasting Proof of Work (PoW) tasks with computations that deliver real-world value. In this case, GPU cycles powering AI model training and inference simultaneously secure the blockchain and mine Pearl’s native cryptocurrency, ¶PRL. This dual utility not only offsets operational costs but also makes AI workloads economically productive.
The discounted endpoint for Gemma-4-31B-it-pearl, an instruction-tuned large language model, is the first product in Together AI’s Pearl-integrated portfolio. According to Omri Weinstein, Co-founder and CEO of Pearl Research Labs, this offering is a milestone in the application of PoUW, demonstrating how AI workloads can generate tangible financial benefits. Weinstein stated, “Pearl changes the unit economics of AI by allowing every GPU cycle to produce a native proof-of-work digital asset, ¶PRL, at no additional cost.”
Proof of Useful Work has been gaining traction as a more efficient and environmentally friendly alternative to traditional blockchain consensus mechanisms. Unlike PoW, which relies on solving arbitrary cryptographic puzzles, PoUW directs computational resources toward valuable tasks such as model training, scientific simulations, or data analysis. These computations are validated through cryptographic proofs, ensuring they meet both blockchain security standards and external utility benchmarks. Recent developments in zero-knowledge proofs and verifiable computing have further pushed PoUW into mainstream adoption, particularly within decentralized AI infrastructure projects.
This collaboration isn’t just a technical innovation—it’s also a strategic move in an increasingly competitive AI space where cost reduction is key. By integrating blockchain rewards into its pricing model, Together AI is betting on the long-term viability of crypto incentives to drive adoption. The timing aligns with a broader trend: decentralized compute networks have been rapidly adopting PoUW to make AI workloads more transparent, efficient, and economically viable. This partnership could serve as a test case for whether such models can scale effectively in commercial AI deployments.
For developers and enterprises interested in experimenting with this technology, the Gemma-4-31B-it-pearl endpoint is now available via Together AI’s platform. Meanwhile, Pearl Research Labs is continuing its efforts to expand the utility and adoption of ¶PRL by collaborating with other AI providers and pushing for standardization in PoUW frameworks.
As the AI and blockchain industries converge, projects like this one highlight how distributed ledger technology can be more than just a database—it can fundamentally reshape the economics of computation. Whether this model gains widespread adoption will depend on its ability to deliver consistent cost savings and seamless integration for businesses navigating the high-stakes world of AI deployment.
Image source: Shutterstock.jpg)