He Built the World's #1 Open-Source Coding Agent

OpenCode Scale & Growth Metrics

• The company reached 4.6 million weekly active users and 13 million monthly active users as of June, representing a 20x increase from the beginning of the year.
• OpenCode processes approximately 7 trillion tokens per day, exceeding the daily volume of OpenRouter (6 trillion).
• Subscription revenue (OpenCode Go) launched in late February reached approximately 160,000 monthly subscribers.
• Annualized revenue from subscriptions is estimated at $18 million, while extrapolated inference revenue based on June data ranges from $31–$33 million.
• Revenue projections based on the most recent week's data suggest annualized inference revenue could reach $38–$40 million within eight months of launch.

Growth Drivers & Strategic Inflection Points

• OpenCode's growth accelerated after Anthropic attempted to block usage of open models within Claude Code subscriptions, inadvertently positioning OpenCode as a viable alternative.
• The strategy of "betting the field" by supporting over 70 open and frontier models allowed users to bypass vendor lock-in, attracting developers seeking flexibility.
• The product serves as a "neutral harness" for enterprise customers, enabling organizations to adopt AI agents without committing to a specific model vendor's infrastructure.
• OpenCode functions as the largest customer for most open-source model labs, creating a symbiotic relationship where OpenCode's token volume drives inference demand for these models.

Global Adoption & Geographic Trends

• China accounts for 17% of traffic, with a significant portion utilizing domestic models like DeepSeek and GLM.
• Developing nations, including Indonesia (4% of traffic) and Brazil (5% of traffic), show high adoption rates driven by the inability to afford $200/month Claude Code subscriptions.
• The U.S. market is also growing rapidly, reflecting a shift toward cost-conscious token budgeting rather than purely unlimited usage.
• Enterprise adoption is characterized by "bottom-up" viral growth, where teams request official security agreements rather than following traditional procurement processes.

Model Usage & Economics

• Data shows DeepSeek Flash is the most used model by token volume, followed by GLM and DeepSeek Pro, often serving as a cost-optimized alternative to frontier models.
• A shift occurred in February where Kimi 2.5 usage surpassed Anthropic's models (Sonnet + Opus) for a four-week span, validating the viability of open-source models for "real work."
• Users employ a "tiered strategy," switching to cheaper models (e.g., DeepSeek Flash) when approaching daily limits to maintain workflow efficiency.
• Unit economics rely on volume discounts for tokens and a mix of free-tier acquisition (subsidized by the company) and high-volume "whale" users paying per token.
• 24-hour GPU utilization is stabilized by the company's global user base, ensuring consistent load across time zones and improving operational efficiency.

Product Design & Historical Context

• The company originated in 2010 (a 16-year-old legal entity) and was rejected by Y Combinator nine times between 2016 and 2021 before being accepted with a serverless platform idea.
• Founders Jay Poldinger and Frank built prior terminal UIs, including a niche "coffee ordering over SSH" product, which established their expertise in terminal-centric developer tools.
• OpenCode was intentionally designed to replicate the experience of modern terminal UIs (like NeoVim) rather than the graphical interface of competitors, targeting hardcore developers.
• The name "OpenCode" and the support for 70+ models were deliberate choices to create an open alternative in a market dominated by two major players (Claude Code and CodeX).

Enterprise Engagement

• Companies like Ramp have embedded the OpenCode agent loop into internal tools, such as Slack bots, bypassing the main UI entirely.
• Enterprises are increasingly asking for granular token management features to limit access to expensive frontier models within different organizational divisions.
• A minority of users are requesting detailed visibility into team usage for internal governance, presenting a potential new feature direction.
• The company reports product-market fit validation when large enterprises proactively contact them to sign security agreements without prior sales outreach.