Skip to main content
OpenSourceAlternatives

Pocket Network

Open source · OSI-approvedStatus: unknownNew

Open network where node operators earn per data request served

Pocket Network is a decentralized, permissionless network where developers publish data services and independent operators earn payment for every request they serve. It suits crypto and multi-chain application teams that want RPC and data access across more than 60 networks without depending on a single provider, and operators who want to monetize spare backend capacity. Since the Shannon mainnet upgrade in June 2025, the protocol runs as an open Cosmos SDK appchain with usage-based pricing instead of subscriptions.[1][2]

Quick verdict

Multi-chain application teams that want decentralized RPC without single-provider dependence benefit most from Pocket Network. Its permissionless relays across 60+ networks, RelayMiner supplier earning, gateway aggregation, and per-request compute-unit pricing cover both consuming and supplying data services, with MIT-licensed Shannon protocol code from June 2025. The material constraints are evidence and maturity: no independent Reddit or G2 firsthand reports surfaced, supplier setup means staking POKT and running server infrastructure, and the stack is still evolving. If managed SLAs matter more than decentralization, compare hosted RPC providers first.[1][2][3][4]

Key features

  • Permissionless relays across 60+ networks Anyone can supply or consume API services with no allowlists or gatekeepers, across more than 60 networks including Solana, Ethereum, and Cosmos chains. Open access is the network's core distinction from allowlisted RPC providers.[1][2]

  • Shannon protocol on the Cosmos SDK The Morse-to-Shannon mainnet transition completed on June 3, 2025 as a full protocol rebuild to a Cosmos SDK and CometBFT stack. The Shannon implementation is open source and MIT licensed, with the catalogue tracking the pokt-network/poktroll repository at v0.1.35.[1][2]

  • RelayMiner supplier software Suppliers run RelayMiner server software alongside Postgres, backend data nodes, and a domain, and the software claims onchain rewards for every relay the supplier's staked service serves. Official step-by-step setup guides cover popular chains such as Ethereum, Solana, and Cosmos.[1][2]

  • Pay per request in compute units Access costs follow usage instead of a flat subscription: applications pay per relay in compute units that carry a fixed dollar value. Gateways can add their own pricing on top, and developers can also publish data services that earn from relay traffic and portal subscriptions.[1]

  • Gateway aggregation for applications Gateways aggregate supplier capacity into application-ready endpoints, with Nodies, Grove, Liquify, Chainstack, and Vulkan listed on the official site. Teams can also start or join a gateway rather than building directly on the protocol.[1]

  • Full data services beyond RPC Shannon expanded the network from limited relays to full data services, including APIs, ETL pipelines, and AI model training data transformation. Developers publish data services that agents and applications pay for per request.[1][2]

  • Stake-backed crypto-economic security Suppliers stake POKT to guarantee honest service, and the network reaches consensus through checkpoints roughly every four blocks, about fifteen minutes. Payments for served relays settle onchain in POKT under the protocol's buy-and-burn and deflationary design.[1]

  • MIT-licensed, DAO-governed stack No license is needed to build on the protocol, and the Pocket DAO decides its direction through forum proposals voted on by POKT holders. The official site reports a community of more than 30,000 members.[1]

Who will love it

  • Multi-chain teams avoiding lock-in Application teams that need RPC access across many chains and want no single-provider dependency fit best: permissionless supply across 60+ networks is exactly the redundancy story Pocket sells.[1][2]

  • Operators monetizing backend capacity Infrastructure operators with spare server capacity and the ability to stake POKT fit the supplier role, earning per relay served through RelayMiner with documented setup guides.[1][2]

  • Open-source and DePIN-aligned projects Teams whose architecture or values favor decentralized physical infrastructure and MIT-licensed open protocols will love the DAO governance and no-license-needed building terms.[1]

Who should avoid it

  • Supplying is heavier than consuming Joining as a supplier means staking POKT, running RelayMiner with Postgres and backend nodes, and surviving a proof-of-service trial period. Teams that only need RPC access should stay on the gateway consumer side rather than supplying.[1][2]

  • Usage bills scale with traffic Per-request pricing means high-traffic applications must model compute-unit spend against flat-rate RPC plans. The fixed dollar value of compute units helps, but gateway markups and exact stake economics were not published on the pages reviewed.[1]

  • Young, still-evolving stack The Morse-to-Shannon transition completed on June 3, 2025, and the RelayMiner documentation describes itself as bleeding edge and still in development. Suppliers should expect evolving setup instructions and changing tooling rather than a settled mature platform.[1][2]

Alternatives and compatibility

EVM and non-EVM chain coverage Supported networks span EVM chains such as Ethereum and non-EVM chains such as Solana and Cosmos, with more than 60 networks available and 30 or more full-data services staked.[1][2]

Standard gateway endpoints Applications consume the network through standard JSON-RPC endpoints such as eth.api.pocket.network, with no App ID or API key on the public portal, and existing ethers.js, web3.py, or viem integrations carry over with a URL change. Gateway providers including Nodies, Grove, Liquify, Chainstack, and Vulkan aggregate supply, and high-volume teams can use Foundation Partnership endpoints or deploy their own PATH gateway.[1]

Cosmos SDK native for contributors Protocol-level contributors work with a Cosmos SDK and CometBFT stack, and the implementation is open source under MIT with public documentation and setup guides.[1][2]

Documented supplier hardware guidance Supplier operators get documented hardware requirements and per-chain setup guides covering the RelayMiner, Postgres, backend node, and domain components of a serving stack.[1][2]

Questions worth asking

What is the best Pocket Network alternative for non-technical users?

A managed RPC provider or a gateway account fits better when nobody on the team wants to touch blockchain infrastructure. Pocket's consumer side is already low-friction, since apps buy through gateways such as Grove, Nodies, or Liquify, but running a supplier means staking POKT and operating RelayMiner servers. Non-technical teams should stay on the gateway consumer side or compare fully managed RPC services first.[1][2]

Can I use Pocket Network without writing code?

There is no documented no-code path to earning as a supplier: operating RelayMiner, a backend data node, Postgres, and a domain is infrastructure work with documented hardware requirements. On the consumer side, using the network needs only a gateway endpoint in the application, which is configuration rather than code. Teams without ops capacity should consume through a gateway instead of supplying.[1][2]

How much does Pocket Network cost per month?

There is no flat monthly subscription published on the official pages. Applications pay per use in compute units, which carry a fixed dollar value, so spend scales with relay traffic. Suppliers face a different cost side: a POKT stake plus the servers to run RelayMiner and backend nodes, with exact stake sizes and gateway list prices not published on the pages reviewed in October 2026.[1][2]

Can I self-host Pocket Network infrastructure?

Yes, on the supply side. Suppliers self-host the RelayMiner operator software alongside Postgres, their own backend data nodes, and a domain, following the official step-by-step chain setup guides and hardware recommendations. Consumers do not self-host anything: they call a gateway endpoint. The RelayMiner documentation describes itself as bleeding edge and still in development, so suppliers should expect evolving setup instructions.[1][2]

Sources and editorial review

Source research. Reviewed on October 9, 2026. Written by Sanjib Sah. AI assisted; reviewed by a human editor.

We have not run or tested this product.

  1. Pocket Network homepage: permissionless API infrastructure (official)

    official · Observed October 9, 2026

    Official homepage: decentralized network where developers publish data services and operators earn per request; Shannon built with Cosmos SDK; 60+ networks; compute units with fixed dollar value; RelayMiner supplier software; gateways Nodies, Grove, Liquify, Chainstack, Vulkan; POKT staking, ~4-block checkpoints, DAO governance, MIT license.

  2. Pocket Network supplier documentation (official)

    official · Observed October 9, 2026

    Official supplier documentation: permissionless supply with no gatekeepers; stake POKT to start earning; RelayMiner plus Postgres plus backend plus domain; proof-of-service trial period; permissionless relays, gateway demand, compute-unit cost for apps; start or join a gateway.

  3. Pocket Network Shannon upgrade retrospective (official docs)

    official · Observed October 9, 2026

    Official Shannon upgrade retrospective: Morse-to-Shannon mainnet transition completed June 3, 2025 as a full protocol rebuild; developer endpoints moved to {CHAIN_ID}.api.pocket.network with standard JSON-RPC and no App ID; suppliers run pocketd plus RelayMiner with Cosmos SDK keypairs; high-volume apps can use Foundation Partnership endpoints or deploy their own PATH gateway.

  4. Pocket Network RelayMiner README (official)

    official · Observed October 9, 2026

    Official RelayMiner README: high-availability supplier implementation that claims rewards for every relay a supplier's staked service serves; step-by-step chain setup guides; documentation explicitly bleeding edge and implementation in development with documented hardware requirements.

Compared with these products

No explicit proprietary baseline is recorded.

Similar projects

Related projects with shared categories, technologies, or licensing.

OpenClaw

OpenClawAIOpen

Your own personal AI assistant for messaging, work, and the web

Stars391.7K
Commitjust now
LicenseMIT

OpenClaw is an MIT-licensed open-source personal AI assistant that runs its own gateway and works from WhatsApp, Telegram, Discord, and other connected channels.

Hermes Agent

Hermes AgentAIOpen

Open-source terminal-native AI agent harness from Nous Research

Stars252.6K
Commitjust now
LicenseMIT

Hermes Agent is Nous Research's open-source terminal-native agent harness that plans and executes supervised tasks on the user's own machine.

OpenCode

OpenCodeAIOpen

AI coding agent for terminal, editor, and desktop

Stars202.9K
Commit8h ago
LicenseMIT

OpenCode is an MIT-licensed AI coding agent available through terminal, desktop, and IDE clients. It supports hosted providers and local models. Code handling depends on the selected provider; optional session sharing uploads conversation history to OpenCode's sharing service.

Ollama

OllamaAIOpen

Run open language models locally with one command

Stars179.9K
Commit14h ago
LicenseMIT

Ollama reduced local model hosting from an afternoon of CUDA troubleshooting to a single pull command. It handles model download, quantisation selection, GPU offloading, and memory management, and exposes an OpenAI-compatible endpoint on localhost.

Excalidraw

ExcalidrawOpen

Collaborative whiteboard with a hand-drawn look

Stars130.9K
Commit23h ago
LicenseMIT

Excalidraw's sketchy rendering is not decoration. Diagrams that look provisional invite comment, where a crisp vector diagram signals a decision already made, so it is unusually good at the early part of a design conversation.

llama.cpp

llama.cppAIOpen

Inference engine that runs LLMs on almost any hardware

Stars126.6K
Commit8h ago
LicenseMIT

llama.cpp is the layer most local model tooling is built on. It is a C and C++ inference engine with no Python runtime, aggressive quantisation support, and hand-tuned kernels for a remarkably wide range of hardware.