Analysis
Anthropic's Claude Code continues to dominate the AI coding tool market by usage, despite rising competitive pressure from OpenAI's Codex and open-source, model-agnostic alternatives like OpenCode. Claude Code reached roughly $1 billion in annualized revenue within six months of its May 2025 launch, an extraordinarily fast ramp for a developer tool, and has continued scaling toward a reported figure approaching $8 billion by May 2026.
OpenAI isn't standing still. Sam Altman confirmed Codex had reached 3 million weekly active users as of early April, with token usage growing more than 70% month-over-month -- genuinely strong growth by any standard. OpenAI's GPT-5.6 Sol model, which reached general availability in early July, is available through Codex and represents OpenAI's latest attempt to close the capability gap that's kept many developers on Claude for coding tasks specifically.
“OpenCode, a model-agnostic tool that runs more than 75 different model providers including local models, still leads Claude Code on raw GitHub star count.”
The open-source flank is the most interesting competitive dynamic. OpenCode, a model-agnostic tool that runs more than 75 different model providers including local models, still leads Claude Code on raw GitHub star count. But Claude Code has meaningfully closed that gap since February, suggesting that even developers who value open-source flexibility in principle are gravitating back toward Anthropic's official, polished, Claude-only harness in practice -- a meaningful data point about how much product polish and first-party integration still matter even in a category that theoretically rewards openness and choice.
The stakes here go well beyond a leaderboard. Coding assistants have become one of the highest-revenue, fastest-scaling product categories in applied AI, and IDE-level tooling is notoriously sticky -- developers rarely switch core coding tools once workflows and muscle memory are established. Whoever wins habit formation now is likely to retain disproportionate share for years, which is why Anthropic, OpenAI and the open-source community are all treating this as a category worth fighting hard for rather than one more feature among many.
What to watch: whether GPT-5.6 Sol's general availability narrows Codex's usage gap with Claude Code over the next two quarters, and whether OpenCode's star-count lead ever converts into comparable revenue or enterprise adoption, or stays a developer-mindshare metric that doesn't move the competitive needle.
