CodeBurn vs opencodex
Which should you use? A side-by-side look at what each one does, who it suits, what it costs, how hard it is to set up, and where each falls short.
Open source · MIT
Local tool that tracks what your AI coding tools spend on tokens, broken down by model, project, and task.
Open source · MIT
A local proxy that lets Codex and Claude Code use models from other providers.
The short answer
Choose CodeBurn if you need…
- Developers who use several AI coding tools and want one view of spending
- People on a Claude subscription who want to see how much of their plan they have used
- Developers who want to find config bloat, such as unused MCP servers or oversized CLAUDE.md files
- Anyone comparing whether a workflow or model change actually lowered cost
Choose opencodex if you need…
- Developers who already use Codex or Claude Code and want to try other model providers
- People who want one local dashboard to manage multiple provider keys and models
- Users who want to route different sessions to different accounts or models
- Anyone running local models through Ollama who wants them inside a coding tool
Side-by-side
| CodeBurn | opencodex | |
|---|---|---|
| Type | Dev tool | Dev tool |
| Price | Free and open source | Free and open source |
| Licence | MIT | MIT |
| GitHub stars | 11.7k | 18.2k |
| Setup difficulty | Beginner | Intermediate |
| How you get it | Install with a package manager | Install with a package manager |
| Main language | TypeScript | TypeScript |
| Last code update | 4 days ago | 4 days ago |
What each one is
CodeBurn
CodeBurn reads the session files that AI coding tools such as Claude Code, Codex, and Cursor write to your disk, then prices every token and shows what you spent. It breaks the total down by tool, model, project, and task, so you can see where a monthly bill actually goes. It runs locally and the README says no account or sign-up is needed.
opencodex
opencodex runs on your computer as a local gateway between AI coding tools and model providers. It accepts requests in the Responses, Chat Completions and Anthropic Messages formats and translates them into what each provider expects. The README says this lets tools like Codex and Claude Code work with models such as Claude, Gemini, Grok, DeepSeek and Ollama.
Best for
CodeBurn
- Developers who use several AI coding tools and want one view of spending
- People on a Claude subscription who want to see how much of their plan they have used
- Developers who want to find config bloat, such as unused MCP servers or oversized CLAUDE.md files
- Anyone comparing whether a workflow or model change actually lowered cost
opencodex
- Developers who already use Codex or Claude Code and want to try other model providers
- People who want one local dashboard to manage multiple provider keys and models
- Users who want to route different sessions to different accounts or models
- Anyone running local models through Ollama who wants them inside a coding tool
Strengths
CodeBurn
- Runs locally with no account, and the README says it reads files already on your disk
- Covers many tools and breaks spend down by model, project, git branch, and task
- Optimize and undo commands back up files before changing them, which makes config edits reversible
opencodex
- Works with many clients and providers through one local address, http://localhost:10100
- Has a web dashboard for adding providers, picking models and managing accounts
- MIT licensed and open source on GitHub
Not ideal for
CodeBurn
- People who want a hosted cloud dashboard across a team, since the README describes it as local
- Users of tools outside the supported list, because it only reads the session files of tools it integrates with
- Anyone needing exact billing reconciliation, since the README presents figures as estimates from session data and does not claim they match invoices
opencodex
- People who only use one provider and do not need routing, since the proxy adds a layer to set up and maintain
- Users who need guaranteed feature parity; the README says streaming, tool calls, reasoning and images work only where the client, model and provider support them
- Non-technical users who are uncomfortable installing Node packages or editing tool configs
Trade-offs
CodeBurn
- Depends on each tool's session file format, so coverage can change when those tools update
- The README describes several surfaces (desktop app, menu bar, Capacity Dock) that appear to be macOS-focused, and the Windows and Linux feature set is not clearly stated
- README is truncated, so details on privacy of the desktop app and full command list are not verified here
opencodex
- The README describes support as conditional on client, model and provider, so behavior varies by combination
- Several setup paths (CLI, desktop app, managed integrations) mean more to learn; the README is long and partly truncated
- Desktop app is labeled beta in the README
Getting started
CodeBurn
Run without installing
npx codeburnInstall globally with npm
npm install -g codeburnopencodex
Install the CLI globally
npm install -g @bitkyc08/opencodexStart the proxy and dashboard
ocx startFrequently asked questions
What is the difference between CodeBurn and opencodex?
CodeBurn: Local tool that tracks what your AI coding tools spend on tokens, broken down by model, project, and task. opencodex: A local proxy that lets Codex and Claude Code use models from other providers. CodeBurn suits developers who use several ai coding tools and want one view of spending, while opencodex suits developers who already use codex or claude code and want to try other model providers.
Are CodeBurn and opencodex free?
CodeBurn: Free and open source. opencodex: Free and open source.
Which is easier to set up, CodeBurn or opencodex?
CodeBurn is easier to get started with (rated beginner); opencodex is rated intermediate and usually needs more setup.
Is CodeBurn or opencodex more popular?
opencodex has more GitHub stars (18.2k) than CodeBurn (11.7k). Popularity isn't the same as fit — check which one matches your use case above.
What are other alternatives to CodeBurn and opencodex?
Other options people compare them with include codebase-memory-mcp, code-review-graph, ECC, Agentic Awesome Skills, ccstatusline.
Keep comparing
Built from our catalogue entries for both products: open-source details come from each project's README (AI-assisted, refreshed weekly) and GitHub data; AI-tool details are editorial. Spotted something wrong? Let us know from either product's page.