The first Pearl-powered endpoint.
- >25%
- Discount on the Pearl-powered endpoint
- 1
- Forward pass, two outputs
- First
- Commercial deployment of the "2-for-1"
Versus the standard Gemma-4-31B-it endpoint. Funded by PRL emissions.
The same matmul that produces tokens for the customer mines PRL for the operator.
Live on a leading neocloud, on a model customers actually use.
Gemma-4-31B-it-pearl is live on Together AI’s inference platform. An instruction-tuned Gemma variant, Pearl-certified end to end, available at the public endpoint today - same OpenAI-compatible API the rest of the catalog uses, priced below the standard Gemma-4-31B-it.
- 400T
- Open-source tokens served / month
- 1M+
- Developers and AI-native apps
- 150MW → 5GW
- Compute, 2026 to 2030
10,000× year-over-year growth, from 30B to 400T monthly tokens.
Cursor, Cognition, Decagon, Ollama, Suno, Cartesia, Bland, and many more.
One of the largest open-AI compute platforms in the world.
Proof-of-Work now runs on top of arbitrary matrix multiplication, with vanishing o(1) overhead. The same forward pass that produces the customer’s tokens produces a Proof-of-Useful-Work attestation that mines PRL. The Pearl plugin drops into the existing pipeline: no retraining, no extra round trips, no second cluster, no added latency. One computation, two outputs.
Input
GPU forward pass
PoUW attestation
Output 01
Inference result
Output 02
NewPRL token
This “2-for-1” technology means the hundreds of billions - and soon trillions - of dollars of compute being deployed for AI workloads will double, for effectively free, to secure Pearl’s Proof-of-Work chain, enabling the most economic inference and training in terms of LLM price-per-token.
“Pearl changes the unit economics of AI, by allowing every GPU cycle powering AI training and inference to simultaneously produce a native proof-of-work digital asset - the PRL coin - at no additional cost. Pearl’s breakthrough technology increases the economic throughput of GPUs and decreases LLM price-per-token.”
