State of FLock Q2 2025

Key Insights

• Circulating market capitalization increased 235% QoQ to $30 million, while FLOCK’s price rose 167% from $0.06 to $0.15.
• Training and delegator nodes increased in Q2 2025, with training nodes up 29% QoQ to 185 and delegators up 21% to 1,280. Validators grew 17% QoQ to 247, while training and validation submissions each declined 52% QoQ.
• Protocol updates included the launch of gmFLOCK, AI Arena v2 with Delegation Pools, and the FLock OFF subnet on Bittensor.
• Partnerships and integrations during Q2 included Base’s Model Context Protocol, Alibaba Cloud’s Qwen, Infini, SpoonOS, and DIMO + Beacon.
• FLock was featured in CB Insights’ AI 100 list and received the IEEE Global Blockchain Conference Best Application Award.

Primer

FLock.io (FLOCK) is a decentralized AI development platform that combines blockchain infrastructure with privacy-preserving federated learning. Its architecture is built around three core components: AI Arena, where models are collaboratively trained; Moonbase, a marketplace for publishing and using models; and FL Alliance, a federated learning framework that coordinates contributors with onchain incentives.

In traditional federated learning, models are sent to local devices for training, and only the updated parameters (not the raw data) are shared back. This preserves privacy but often depends on centralized servers and leaves challenges around incentives and security. FLock addresses these limitations by using blockchain for decentralized coordination, verifiable governance, and transparent incentives, allowing communities to propose, train, and deploy AI models in a trust-minimized way.

This design supports FLock’s broader vision: to democratize the AI lifecycle from data sourcing and model design to training, validation, and deployment. The platform is anchored in ongoing academic research, with multiple peer-reviewed publications in venues such as IEEE journals and NeurIPS workshops. For a full primer on FLock, refer to our Initiation of Coverage report.

Website / X / Discord / Telegram / LinkedIn

Key Metrics

Financial Overview

Market Capitalization

FLOCK’s circulating market capitalization increased 235% in Q2 2025, rising from $9 million to $30 million, while the token price grew 167%, from $0.06 to $0.15. The larger increase in market capitalization relative to price is due to circulating supply expanding over the quarter through staking rewards, ecosystem incentives, or scheduled token releases. The addition of Uniswap liquidity on Base also broadened market accessibility, increasing the share of tokens actively circulating. As a result, both higher valuation per token and a larger circulating supply contributed to the overall market cap growth.

Staking Ratio

Previously, staking FLOCK was required for all participants in AI Arena and was essential for role eligibility, task access, and incentive alignment. Participants in active tasks fell into one of three roles:

• Training Nodes staked FLOCK to develop or fine-tune models using private data, earning rewards based on the quality and ranking of their submissions.
• Validators staked FLOCK to evaluate models using standardized datasets, with validation frequency influenced by stake and subject to rate-limiting.
• Delegators staked FLOCK on behalf of Training Nodes or Validators and received a share of rewards, with payouts time-weighted and dependent on performance and reward-sharing ratios.

Staked tokens functioned as collateral to discourage dishonest behavior. Smart contracts automatically distributed rewards, and slashing penalties applied for protocol violations.

In Q2 2025, FLock introduced gmFLOCK as the staking unit for AI Arena participation. gmFLOCK is a non-transferable token generated by locking FLOCK for a user-selected period of up to 365 days, with longer lockups receiving higher multipliers. Training Nodes, Validators, and Delegators must now convert FLOCK into gmFLOCK to participate and earn rewards. The transition away from the prior staking model contributed to a sharp decline in the overall staking ratio, as previously staked tokens had to be re-locked under the new system.

FLOCK’s staking ratio declined in Q2 2025, falling from 41% at the end of Q1 to 33% by quarter-end. The ratio peaked at 53% on May 15 before a sharp correction in late May brought it back to the low-30% range, where it stabilized through June. The shift reflected weaker staking incentives, and the introduction of gmFLOCK in late April, which added an extra conversion step for participation. Staking rewards decreased from $2.7 million in Q1 to $1.9 million in Q2, reducing the financial appeal of remaining locked. At the same time, FLOCK’s price rose 167% over the quarter, likely encouraging some holders to unstake to realize gains or provide liquidity. The decline in staking increased the amount of liquid supply available in the market, amplifying the impact of price appreciation on market capitalization.

Network Overview

FLock’s network participation is measured through several core metrics:

• Training submissions track the number of model updates contributed by Training Nodes during a round. Each submission reflects local training on private data, and the quality and ranking of updates directly determine reward distribution.
• Validation submissions measure evaluations performed by Validators, benchmarking submitted models against standardized datasets to ensure accuracy and fairness.
• Training Node count represents unique participants actively developing or fine-tuning models.
• Validator count captures participants engaged in evaluating models.
• Delegator count shows how many tokenholders lock FLOCK to generate gmFLOCK, which is then used to delegate support to Training Nodes or Validators. Delegators share in rewards without running infrastructure themselves.
• Validator and Training rewards (USD) reflect the aggregate value of incentives distributed to active roles, serving as a measure of overall economic support for network activity.

Together, these metrics provide a view into FLock’s decentralization, activity levels, and incentive alignment.

Training Activity

Training nodes expanded from 143 at the end of Q1 2025 to 185 by the end of Q2 2025, a 29% QoQ increase. This growth was supported by the introduction of gmFLOCK staking in late April and the AI Arena v2 upgrade in May, which reinforced incentives for node operators and delegators. New training tasks and partnerships, including integrations with Alibaba Cloud’s Qwen, Base’s Model Context Protocol, and application-focused initiatives like Baby4D and the DIMO + Beacon fuel-efficiency model, also broadened participation.

Training submissions grew sharply in Q1 before moderating in Q2. Daily submissions rose from near zero in Q4 2024 to over 4,700 in Q1, then fell to about 2,300 in Q2, a 52% decrease QoQ. Cumulative submissions continued to rise, showing that participation persisted despite the lower daily rate. The moderation reflected more specialized workloads introduced in Q2, which reduced submission frequency but expanded the network’s application scope.

Validator Activity

Validation submissions on FLock followed a similar trajectory to training activity. Daily submissions surged to over 412,000 in Q1 2025 but declined to about 197,000 in Q2, a 52% QoQ decrease. Despite this, cumulative validation submissions kept rising, indicating sustained engagement.

The validator set expanded from 211 to 247 during Q2 (+17%), meaning more participants were active even as the average number of validations per node fell. The moderation in throughput coincided with the rollout of gmFLOCK and Delegation Pools, which shifted incentives toward longer-term staking and away from maximizing short-term submission volume.

The rolling 30-day yield measures average rewards for training nodes and validators over the prior 30 days, smoothing short-term fluctuations to show incentive trends.

In Q1 2025, yields declined as participation scaled up: training node yields fell from 175% on January 1 to 29% by March 31, while validator yields dropped from 1,404% to 14%.

In Q2, yields stabilized in the double-digit range before shifting in late May, when training node yields briefly spiked above 34% and validator yields similarly touched 15%. At quarter-end, they stood at 7.8% and 5.8%, respectively.

Network Dynamics

In Q2 2025, validator and delegator participation continued to expand, though at a slower pace than in Q1. Validators grew from 211 at the end of March to 247 by June (+17% QoQ), compared to an 11x increase in the prior quarter. Delegators rose from 1,060 to 1,280 over the same period (+21% QoQ), following a 22x jump in Q1. The delegator-to-validator ratio remained steady, moving slightly from 5:1 in Q1 to 5.2:1 in Q2, underscoring that delegation remained the preferred form of participation for tokenholders.

Validator growth moderated as the network approached operational stability after the rapid build-out of Q1. By contrast, delegator expansion was supported by the introduction of gmFLOCK staking and Delegation Pools which lowered barriers to entry and reinforced incentives for indirect participation. The persistence of a stable ratio highlights that incentives continued to balance direct and delegated roles within the network.

Qualitative Analysis

Technology and Protocol Developments

• gmFLOCK launch: On April 28, 2025, FLock launched gmFLOCK, a staking-based participation mechanism where locked FLOCK generated gmFLOCK for Training Node Operators, Validators, and Delegators. The system rewarded longer lockups with higher multipliers, encouraging long-term participation.
• FLock OFF subnet launch: On May 2, 2025, FLock deployed FLock OFF, a permissionless federated learning subnet on Bittensor (UID 96) designed for training Small Language Models (SLMs) on edge devices. Season 1 mining began the same day. Like other Bittensor subnets, participants earn rewards in TAO, not FLOCK, and the subnet does not issue a separate token. FLock OFF expands FLock’s federated learning framework into the Bittensor ecosystem, complementing AI Arena by creating an additional venue for model training and experimentation.
• AI Arena v2 upgrade: On May 27, 2025, FLock released AI Arena v2, a major platform upgrade that introduced Delegation Pools to replace Mini Pools. The update integrated gmFLOCK and implemented freeze windows around training periods to prevent last-minute delegations and ensure fairer reward distribution.
• DEX liquidity expansion: During Q2 2025, FLock expanded DEX liquidity by adding Uniswap pools on Base with trading pairs FLOCK/USDT, FLOCK/USDC, and FLOCK/WETH, improving token accessibility across the ecosystem.
• Web3 Foundation Model adoption: On June 5, 2025, Base adopted FLock’s Web3 Foundation Model, replacing Anthropic’s Claude as the onchain execution engine for the Model Context Protocol (MCP), following its stronger performance on Web3 benchmarks.

Partnerships and Ecosystem Expansion

• SpoonOS integration: On April 29, 2025, FLock partnered with SpoonOS, enabling FLock-trained models to function within agentic AI ecosystems via the Model Context Protocol (MCP) and Agent-to-Agent Protocol. This allowed models to interact across interoperable, onchain AI environments.
• Infini Pay-fAI collaboration: On May 23, 2025, FLock announced its collaboration with Infini through the Pay-fAI initiative, which powered InfiniGPT using FLock’s federated learning. The integration was designed to show decentralised AI utility in financial services without centralising sensitive user or transaction data.
• Alibaba Cloud Qwen partnership: On June 2, 2025, FLock announced its partnership with Alibaba Cloud’s Qwen, a Tier 1 open-source LLM ranging from 0.6B–235B parameters. The collaboration focused on joint research in federated learning and privacy-preserving model training, with applications across healthcare, finance, and education.
• DIMO & Beacon Protocol fuel efficiency model: On June 18, 2025, FLock partnered with DIMO and Beacon Protocol to develop a privacy-preserving vehicle optimisation model. The tool used community-owned driving data to provide personalised route optimisation and fuel efficiency insights.

Research and Educational Initiatives

• AI watermarking explainer: On April 6, 2025, FLock published its AI watermarking explainer, outlining watermarking methods, their importance for intellectual property protection, and compliance requirements under the EU AI Act.
• Model Context Protocol (MCP) explainer: On April 28, 2025, FLock released its Model Context Protocol (MCP) explainer, covering MCP’s architecture, lifecycle stages, security considerations, and adoption trends. The post highlighted MCP’s potential role in decentralised AI agent coordination.
• Baby4D tool: On June 18, 2025, FLock introduced Baby4D, a federated AI-powered prenatal tool that generates ultrasound-based facial renderings while preserving privacy.

Community and Talent Development

• Elite Trainer Programme: On May 20, 2025, FLock launched the Elite Trainer Programme, designed to onboard high-performing ML engineers, AI developers, and data scientists as training nodes. Participants gained access to token-gated tasks, earned a 50% revenue share from training rewards, and contributed to the development of decentralised model training.
• UK AI Agent Hackathon: On April 30, 2025, FLock sponsored the UK AI Agent Hackathon at Imperial College London, billed as Europe’s largest student-led agentic AI hackathon. The event gave participants access to FLock’s Web3 Foundation Model, and winning projects included NovaChat-V2 (a blockchain chatbot), Aetheria (a gamified NFT experience), and Crypto QR Wizard (QR-based transaction tool).

Recognition and Industry Presence

• CB Insights AI 100: On June 17, 2025, FLock was included in CB Insights’ AI 100 list of the world’s most promising AI startups. It was the only decentralised AI project featured in the infrastructure and data category.
• IEEE Global Blockchain Conference award: Announced on July 9, 2025, FLock received the Best Application Award at the IEEE Global Blockchain Conference for its work on blockchain-enabled federated learning in healthcare, showing that hospitals could collaboratively train models while preserving patient privacy.
• The Web Conference 2025: In Q2 2025, FLock presented at The Web Conference 2025, becoming the first community-driven AI training platform to be featured. The appearance provided academic and industry recognition for its decentralised architecture and governance model.
• Industry events: In Q2 2025, FLock participated in several industry events, including the Hong Kong Web3 Festival, Token2049 Dubai, Consensus Toronto, Proof of Talk in France, and ETH Seoul. These appearances positioned the project as an active participant in decentralised AI and Web3 discussions.
• UNDP SDG Blockchain Accelerator partnership: In August 2025, after the close of Q2, FLock was selected by the United Nations Development Programme (UNDP) as its Strategic Partner for AI in the SDG Blockchain Accelerator, where it will mentor pilot projects applying federated learning and decentralized AI to real-world development challenges.

Closing Summary

In Q2 2025, FLock’s circulating market capitalization rose 235% to $30 million, outpacing a 167% increase in token price as circulating supply expanded. The staking ratio declined from 41% to 33%, reflecting reduced incentives, profit-taking amid price appreciation, and the transition to gmFLOCK staking. Training nodes grew 29% QoQ to 185, validators increased 17% to 247, and delegators rose 21% to 1,280, while training and validation submissions fell 52% QoQ as workloads shifted toward more specialized tasks.

Strategic developments included the launch of gmFLOCK, AI Arena v2 with Delegation Pools, and the FLock OFF subnet on Bittensor, alongside integrations with Base’s Model Context Protocol, Alibaba Cloud’s Qwen, Infini, SpoonOS, and DIMO + Beacon. Recognition in CB Insights’ AI 100 list, awards at the IEEE Global Blockchain Conference, and participation in major Web3 and AI events elevated FLock’s profile as a decentralized AI platform.

Looking ahead, FLock enters the second half of 2025 with expanded tokenomics, a growing ecosystem, and a stronger industry presence, though participation trends highlight the need to balance incentives as the network scales into more complex, specialized AI workloads.

Let us know what you loved about the report, what may be missing, or share any other feedback by filling out this short form. All responses are subject to our Privacy Policy and Terms of Service.

This report was commissioned by FLock.io. All content was produced independently by the author(s) and does not necessarily reflect the opinions of Messari, Inc. or the organization that requested the report. The commissioning organization may have input on the content of the report, but Messari maintains editorial control over the final report to retain data accuracy and objectivity. Author(s) may hold cryptocurrencies named in this report. This report is meant for informational purposes only. It is not meant to serve as investment advice. You should conduct your own research and consult an independent financial, tax, or legal advisor before making any investment decisions. Past performance of any asset is not indicative of future results. Please see our Terms of Service for more information.

No part of this report may be (a) copied, photocopied, duplicated in any form by any means or (b) redistributed without the prior written consent of Messari®.