What is Dolphin (POD)?

Quick Facts

• Token: POD (originally launched as DPHN, rebranded to POD)
• Blockchain: Base (Coinbase's Layer-2 network)
• Category: AI lab + Decentralized Physical Infrastructure (DePIN)
• Core function: Aggregates idle GPU compute for AI inference
• Revenue model: 100% of protocol revenue used to buy back POD tokens
• Key partner: Venice.AI — Dolphin supplies its default uncensored model
• Governance: DAO structure with no traditional equity

Introduction

Dolphin (POD) is the native token of the Dolphin Network, an AI-focused DePIN project built on Base. The project combines an AI research lab with a distributed GPU inference network, allowing anyone with idle GPU hardware to earn POD by serving AI model requests.

The token name POD comes from 'a pod of dolphins,' tying directly back to the project's branding and identity.

History & Background

Dolphin began as an AI laboratory focused on fine-tuning open-source models — such as Meta's Llama and Alibaba's Qwen — removing built-in refusals to create what the team calls uncensored models. These models accumulated over five million monthly downloads on Hugging Face, establishing Dolphin as a recognized name in open-model development.

The project originally launched its token under the symbol DPHN before migrating to a new contract and rebranding to POD in early 2026. Dolphin subsequently became the default model provider on the Venice.AI platform.

How Dolphin Works

Dolphin uses a peer-to-pool architecture. Instead of pairing a buyer directly with a single GPU provider, inference requests are routed across a shared pool of GPUs contributed by network operators. Each AI model runs distributed across multiple GPUs in this pool.

Node operators must bond POD tokens to participate, creating a staking-and-slashing mechanism that enforces honest behavior. The network uses live-weight proofs and logprob fingerprinting to verify that nodes are genuinely running the correct models.

Tokenomics

POD functions as both the rewards currency and the economic flywheel of the network. Node operators earn POD for processing inference requests. On the demand side, all revenue received by the protocol is used to repurchase POD from the open market — directly offsetting the token emissions paid to operators.

Operators can stake their POD into the xPOD vault to earn additional yield. A reward multiplier system, inspired by Curve Finance's LP boost model, lets bonded operators earn proportionally more based on their staked balance.

Ecosystem & Use Cases

• Inference network: GPU owners contribute compute and earn POD rewards.
• AI model access: Developers and applications access Dolphin's uncensored models via the network's API.
• Venice.AI integration: Dolphin serves as the core default uncensored model provider for Venice, a privacy-first generative AI platform.
• Staking: POD holders can stake into xPOD to earn a share of liquid claim fees and boost node rewards.

Team, Governance & Community

The team operates under a DAO structure with no equity. The project has stated that any future legal entity would be structured as a non-profit, similar to Morpho's model. This approach places token holders at the center of the network's ownership and direction.

The community is active on X (formerly Twitter) under @dphnAI, with additional channels on Telegram and YouTube.

Advantages

• Revenue-to-buyback model: 100% of protocol income flows directly into POD buybacks, aligning token value with network usage.
• Established AI credentials: Millions of Hugging Face downloads demonstrate real-world traction before the token launch.
• Venice.AI partnership: A high-profile anchor customer provides immediate, verifiable demand for the network.
• Staking incentives: xPOD staking and bonding give long-term holders a path to compounding yield.

Risks & Challenges

• Smart contract risk: Third-party audits have flagged that the contract creator retains certain administrative capabilities, including fee changes and minting.
• Early-stage network: The supply side launched first; public API access and full demand-side scale are still developing.
• Token concentration: A significant portion of tokens is held in a small number of addresses, which can increase volatility.
• Competitive landscape: Decentralized AI inference is a rapidly growing sector with well-funded competitors.

Long-Term Vision

Dolphin aims to become a foundational layer for decentralized AI inference, making open and uncensored model serving accessible without relying on centralized cloud providers. By anchoring token value to real inference revenue rather than speculation, the project positions POD as a utility-first asset that grows alongside genuine AI demand on-chain.