Harness Hub

The directory · side by side

Compare

Line up two to four entries and see where they differ. The link in your address bar shares this exact comparison.

Codex vs opencode. Rows marked Differs have different values.

Open source

Codex

Remove

Open source

opencode

Remove

TypeOpen sourceOpen source
Categories
In one lineOpenAI's terminal coding agent.The open source coding agent — terminal-first, model-agnostic.
Stage & funding——
GitHub starsDiffers≈131k≈223k
LicenseDiffersApache-2.0MIT
LanguageDiffersRustTypeScript
Last pushOct 28, 2026Oct 28, 2026
Works with——
Built onDiffers—
Built on itDiffers—
Similar toDiffers
People linkedDiffers24
EventsDiffers0
Pros
  • DocsApache-2.0 open-source CLI that sandboxes by default (workspace-write, using Seatbelt on macOS, bubblewrap on Linux and a native Windows sandbox).[1][2]
  • DocsIncluded in every ChatGPT plan from Free to Enterprise, and Pro currently has no five-hour limit (weekly limits may still apply).[1]
  • User reportsUsers who switched from Claude report comparable coding quality, better value and faster responses on Plus.[1][2]
  • MetadataVery fast release cadence (rust-v0.165.0 on Oct 27, 2026; about 132k stars).[1]
  • User reportsWorks with any provider or model, with LSP integration and configurable subagents that can each use a different model.[1][2]
  • User reportsUsers measure its prompt overhead as much leaner than Claude Code's (about 7k vs 33k tokens), and it has a plugin ecosystem.[1][2]
  • User reportsA popular fallback and local-model harness when subscription limits run out; 2.0 added background tasks.[1][2]
  • MetadataMIT-licensed, about 223k stars, with near-daily releases (v1.19.5 on Oct 28, 2026).[1]
Cons
  • User reportsUsage limits are opaque and shifting; some users hit the five-hour cap on the $20 plan during a basic task.[1][2]
  • User reportsHeavy default context: a Pi user says Codex loads close to 20k tokens of "cruft" before your first message.[1]
  • MetadataAbout 25k open GitHub issues and PRs, and critics point to a bloated Rust dependency tree (109 duplicate transitive deps; 75-minute release builds on a laptop).[1][2][3]
  • User reportsUsers report quality and usage swings after model launches.[1][2]
  • User reportsIn October 2026 the OpenCode team disclosed unauthorized access to a legacy system that may have exposed BYOK provider keys saved in hosted workspaces.[1][2]
  • User reportsUsers question its security posture: no built-in sandbox, and title generation sent prompts through the Zen gateway by default.[1][2]
  • User reportsUsers say 2.0 moved tool calls into a background process, which silently broke their sandbox wrappers, and that RAM use is high.[1][2]
  • MetadataThousands of open issues and PRs, and some users get better results from other harnesses with the same model.[1][2]
Best forChatGPT subscribers who want an open-source, sandboxed-by-default coding agent bundled with their plan.Developers who want a model-agnostic terminal agent with low prompt overhead and per-subagent model choice.
Watch out forPlan for opaque, changing usage quotas, and trim the default context if token efficiency matters.Run it inside your own sandbox and set your own small model for auxiliary calls, so prompts don't leave your machine by default.
Links

Popular comparisons