Felan vs. OpenAI Codex
Both run coding agents in the cloud from schedules and GitHub events. Codex Cloud runs OpenAI's own harness on OpenAI models. Felan runs models from every provider and can run the whole platform on your infrastructure.
At a Glance
Agent
Felan: Open-source Felan Code, tuned for token cost
OpenAI Codex: Codex CLI, Apache-2.0; Codex Cloud harness is closed
Models
Felan: All major providers, plus local and private models
OpenAI Codex: OpenAI models in the cloud; other models in the local CLI only
Your own keys
Felan: Bring your own keys or subscription
OpenAI Codex: OpenAI API key in the CLI, SDK, and IDE only; no cloud features
Triggers
Felan: Events, schedules, or by hand
OpenAI Codex: Schedules, app events, or GitHub Action events
Self-hosting
Felan: Whole platform on your infrastructure, on Enterprise
OpenAI Codex: Only the executor, built against the Agents API; OpenAI keeps the harness
Team plan
Felan: €99 per month, unlimited members; models billed separately
OpenAI Codex: $20 per user per month, 2+ users, billed annually
Where Felan Differs
Models From Every Provider
Codex Cloud runs OpenAI models. Other providers and local models are a local CLI feature, not available for background or cloud work. Felan runs every major provider, plus local and private models, in the background.
- Anthropic
- OpenAI
- DeepSeek
- Moonshot AI
- Local model
- Private endpoint
The Whole Platform on Your Servers
OpenAI's self-hosted option runs only the executor in your environment; the Codex harness and model calls stay on OpenAI's infrastructure. Felan Enterprise runs the whole platform on your infrastructure.
Your infrastructure
- Agent loop
- Tools
- Repos
- Model endpoint
One Price, Not per Seat
Codex through ChatGPT Business costs $20 per user each month. Felan Team is €99 a month for unlimited members, with models billed at provider rates.
- Members
- Whole team
- Model tokens
- Provider rate
- Sign-up
- Self-serve
- Team plan
- €99 / month
Common Questions
Is Felan an OpenAI Codex alternative?
Yes, for teams that want background agents on models from more than one provider, or the whole platform on their own infrastructure. Both trigger from schedules and repository events and return pull requests for review.
Can Codex use models other than OpenAI's?
Not in the cloud. Codex Cloud and the GitHub Action run OpenAI models. The local Codex CLI can use custom providers or local models with Ollama or LM Studio, but that is a local, not a background, capability. Felan runs any major provider, plus local and private models, in the background.
Can I self-host OpenAI Codex?
Partly, and only by building against OpenAI's Agents API: you run the executor in your own environment, but the Codex harness and model calls always run on OpenAI's infrastructure. Felan Enterprise runs the whole platform, including the agent, on your infrastructure.
How does Felan pricing compare with Codex?
ChatGPT Business is $20 per user per month billed annually (2 or more users), and Codex shares that plan's usage. Felan Team is €99 a month for the whole team, and model usage is billed by your provider or through Felan credits at provider rates.
Is the Codex CLI open source?
Yes, under Apache-2.0. That license covers the local CLI repository; it does not cover Codex Cloud's hosted harness. Felan Code, the agent Felan's background agents run, is open source too.
When is Codex a better fit?
When your team already pays for ChatGPT and wants Codex bundled with it, or when developers mainly want a local CLI agent with OS-level sandboxing.
Sources: Codex scheduled tasks, Codex pricing, Agents API self-hosted sandboxes, Agents API architecture, Codex advanced configuration, Codex CLI license, Codex GitHub Action.
Keep Your Attention for the Hard Problems
Start with one repo and one recurring task.