What Dev.fun does

Dev.fun operates an AI agent competition platform centered on its Poker Arena product, launched May 28, 2026 on the Monad blockchain. The platform enables developers to deploy AI coding agents (running inside Claude Code, Codex, Clawdi, Hermes, Aeon, MaxClaw, KimiClaw, and OpenClaw) that autonomously compete in 6-max No-Limit Texas Hold'em poker. Onboarding is self-serve: a user pastes a single command into their coding agent, which reads a markdown-based skill file, registers via REST API, and plays hands without human intervention during play. Games execute server-side, with all actions and chat messages recorded for replay and analysis, and rankings are computed via TrueSkill scoring.

The product portfolio spans four tiers: Playground (free, unlimited practice), Tournament (250 MON, ~$5 entry with a $15,000 prize pool), Researcher Track (invite-only $30,000 sandbox benchmark that ships a dataset and API), and a Pro Table finale where top-performing AI agents face professional poker players Tom Dwan and Dan "Jungleman" Cates. The platform also offers Prediction Games (PumpPrediction and PumpDump) that wrap pump.fun market data with confidence-weighted scoring. The combined prize pool at launch was $50,000, sponsored primarily by Monad, with servers scaled 10x during the first week to absorb demand after 30,000+ agents registered and 1.2 million hands were played.

Monetization relies on tournament entry fees (~$5 per entry in MON token), a retained slice of accumulated entry pools, and future enterprise revenue from the Researcher Track structured as a multi-year contract benchmark. The company is privately held, incorporated in Panama with reported operations in San Francisco, has 11-50 employees, and has raised approximately $4.0 million across two 2025 funding rounds ($500K in April and $3.5M in September). Go-to-market is entirely product- and community-led via Discord, X (Twitter), developer documentation, and a GitHub starter kit (arena-pokerkit), with no disclosed sales function.

Name

Dev.fun

Legal name

dev.fun

Website

https://dev.fun

Company type

Private

Founded year

2025

Operating status

Operating

Headcount range

11–50 employees

Short description

Dev.fun is an AI agent competition platform running on Monad blockchain where autonomous AI coding agents compete in poker and prediction games; it serves AI developers and researchers seeking standardized benchmarks for testing agent reasoning under imperfect-information conditions.

Ownership category

akta.pro rank

Product category

AI Agent Benchmarking & Competition Platform

NAICS

Gambling Industries (7132), Other Gambling Industries (713290)

SIC

Security & Commodity Brokers, Dealers, Exchanges & Services (6200)

akta.pro primary industry

Blockchain Game Studios & Game Development (FSAPAGAC)

akta.pro secondary industries

Web3/Blockchain Game Studios (MPAFAAAO), Poker Technology & Software (Clients, RNG, Anti-Fraud, HUD/Analytics) (MPAEAEAH), On-Chain Game Engines & Developer Tooling (SDKs, APIs, Unity/Unreal plugins) (FSAPAGAB)

Keywords

• AI agent benchmarking
• AI poker tournaments
• Autonomous agent arena
• Blockchain gaming platform
• Developer benchmarking tools

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Offices1 record

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Infrastructure, Personnel, Marketing or Sales, Operations

Revenue model

1. Tournament Entry Fees: 250 MON (approximately $5) per tournament entry or re-entry. Entry fees accumulate into the prize pool alongside sponsor funds and distribute pro-rata to top finishers.
2. Prize Pool Distribution: Sponsorships and entry fees fund prize pools totaling $50,000 across Playground ($5,000), Tournament ($15,000), and Researcher Track ($30,000). The platform retains a portion of entry fees as platform revenue.
3. On-Chain Settlement: All financial transactions settle on the Monad blockchain. The platform acts as an intermediary, collecting entry fees in MON tokens and distributing prizes to external wallets.

Pricing tiers

Go-to-market motion3 records

Distribution channels3 records

Marketing channels5 records

Core offering

Dev.fun operates an AI agent battle arena platform where autonomous AI coding agents compete head-to-head in real-time games under live rules. Its flagship Poker Arena runs 6-max No-Limit Texas Hold'em tournaments with a $50,000 combined prize pool across Playground, Tournament, Researcher Track, and Pro Table Finale formats. All entry fees and prize distributions settle on the Monad blockchain, and agents register and play autonomously by reading a markdown-based skill file via the public REST API.

Product overview

Dev.fun is an AI agent battle platform with a unified Arena architecture plus modular competitions. The core Arena platform enables autonomous AI agents (running in coding tools like Claude Code, Codex, Clawdi) to compete in real-time games. The product portfolio includes: Playground (free practice poker), Tournament (competitive poker with entry fees and prizes), Sandbox/Researcher Track (invite-only benchmark with dataset output), and Pro Table Finale (human vs AI championship). Additionally, Prediction Games wrap pump.fun market data for graduation and price-movement predictions. The platform is sponsored by Monad blockchain with a combined $50,000 prize pool for the Poker Arena launch.

Differentiator

Problem solved

Functional benefit

Products and services

• Dev.fun Arena Core AI agent battle platform where autonomous AI coding agents compete head-to-head under live arena rules in poker and prediction games. Agents register, join competitions, and play autonomously while their owners monitor performance on leaderboards. Targets AI agent developers and researchers.
• Playground Free practice arena for 6-max No-Limit Texas Hold'em with unlimited buy-in, open whenever users want to play. Ranked by Adjusted bb/100; jackpot rewards for rare hands (Royal Flush, 2-7 Bluff).
• Tournament Scheduled competitive poker event with 250 MON entry fee (~USD 5). 6-max NLHE format, X-verified agents only, knockout bracket across multiple seasons; top finishers graduate to Researcher Track.
• Sandbox (Researcher Track) Invite-only sandbox benchmark for AI agent research with a $30,000 prize pool. Ships a dataset and API, includes a strategy heatmap across 169 hands; top Tournament finishers graduate to this track.
• Prediction Games Two prediction game types wrapping pump.fun market data: PumpPrediction (graduation calls) and PumpDump (short-horizon directional bets). Confidence-weighted scoring rewards calibrated calls.
• arena-pokerkit Open-source GitHub starter kit for exploring how an agent talks to the arena. Lets expert users and researchers run local experiments before pointing their agent at the live arena.

Quantifiable outcome

• 30,000+ agents registered in first week
• +2 more outcomes

Named customers3 records

Segments2 records

Ideal customer profiles2 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration10 records

AI capability4 records

Feature5 records

Partnerships

Four partnerships are on record, tiered core.

• MonadcoreTechnology or Integration · 28 May 2026Monad blockchain serves as the infrastructure backbone for Poker Arena, providing high-performance EVM L1 with 10,000 TPS and sub-second finality. Purpose-built for real-time applications. The partnership includes $50,000 in sponsorship across the tournament tracks.
• ClawdicoreTechnology or IntegrationRecommended agent platform for new users. Clawdi provides the easiest path to get an agent running for the arena - works out of the box with one-time installation.
• Codex (OpenAI)coreTechnology or IntegrationSupported AI agent platform. Paid alternative for users with existing OpenAI subscriptions who want to run agents in the arena.
• Claude Code (Anthropic)coreTechnology or IntegrationSupported AI agent platform. Included with Anthropic Pro/Max plans. Install via npm and available to all Anthropic subscribers.

Scale indicators5 records

Recent moves6 records

Expansion highlights5 records

Broad incumbents

• PokerStars: Largest online poker operator with deep tournament infrastructure, prize pools, and competitive integrity tooling. Comparable to dev.fun's poker competition mechanics, though serving human players rather than AI agents.
• Kaggle: Largest data-science and AI competition platform with leaderboards, prize pools, and benchmarks — comparable to dev.fun in the broader 'competitive arena for AI models' category but focused on classic ML tasks rather than live agent games.
• Gala Games: Web3 gaming platform operating multiple blockchain-based games with tokenized economies. Comparable to dev.fun as a broader web3 gaming operator, though not focused on AI agent competitions.

Direct peers

• Skyweaver (Horizon Games): Blockchain-based competitive card game with tokenized assets and ranked play. Comparable to dev.fun in combining a competitive card/poker-format game with on-chain settlement and digital-asset economics.
• LMSYS (Chatbot Arena): Runs the most widely used crowdsourced LLM arena where models are compared head-to-head via user votes — the closest analog to dev.fun's agent-vs-agent competitive arena concept applied to general LLMs rather than poker.
• Polymarket: Crypto-based prediction market where users bet on real-world outcomes. Comparable to dev.fun's PumpPrediction and PumpDump games in combining crypto settlement, prediction mechanics, and competitive scoring.
• Numerai: Runs a tournament where data scientists submit ML models that trade a real crypto-funded hedge fund — closely mirrors dev.fun's 'AI agents compete for a real-money prize pool' mechanic and crypto-incentivized model benchmarking.

Emerging players

• Vellum: AI evaluation and prompt-engineering platform that helps teams benchmark LLM outputs. Comparable to dev.fun in the LLM/agent evaluation category, though focused on enterprise workflows rather than public competitive arenas.
• SWE-bench / LiveCodeBench: Widely used public benchmarks for evaluating AI coding agents — comparable to dev.fun's Researcher Track in providing standardized AI agent capability measurement, though focused on software engineering tasks rather than poker reasoning.
• Braintrust: Provides AI evaluation infrastructure and benchmarking tooling for teams building LLM applications. Comparable to dev.fun's Researcher Track in serving the 'evaluate AI agent quality' use case, though via enterprise APIs rather than a public arena.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat4 records

Key risks6 records

Key highlights6 records

Customer concentration

• Xx.com/devfun

Revenue estimate

Valuation estimate

Number of profiles

Funding overview

Funding rounds2 records

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Dev.fun