Both run coding agents in the background. Felan lets you pick the models, pay provider rates with no markup on tokens, and run on our managed cloud or your own infrastructure.
Felan: Open-source Felan Code
Devin: No source license published
Felan: All major providers, plus local and private models
Devin: Hosted model catalog; no local models published
Felan: Bring your own keys or subscription
Devin: No bring-your-own-keys option published
Felan: Provider rates, no markup on tokens
Devin: Credits or ACUs for actions, VM time, and bandwidth
Felan: Managed cloud, or fully self-hosted on Enterprise
Devin: Devin Cloud; Outposts keep the agent loop there
Felan: €99 per month per team, not per seat
Devin: $40 per full seat, $80 monthly minimum, plus credits
Dependency updates and failed-build triage run many times a week, so cost grows with every run.
Devin: Each run draws credits or ACUs metered on Devin's actions, VM time, and bandwidth.
Felan: Each run pays provider rates with no markup, on a cheaper model when the task allows.
Security needs code, prompts, and inference inside your perimeter.
Devin: With Outposts, tools run on your machines, but inference and planning stay in Devin's cloud.
Felan: On Enterprise, the whole agent runs on your infrastructure, with local or private models if you need them.
Your team already pays a model provider directly, or has a subscription.
Devin: Usage is billed through Devin credits; no bring-your-own-keys option is published.
Felan: Bring your own keys or subscription, and agents run against the contract you already have.
Devin offers a catalog of hosted models. Felan also runs local and private models, with your own keys.
Pay your provider directly, or buy Felan credits at provider rates. Devin meters usage in credits or ACUs.
On Enterprise, Felan runs entirely on your infrastructure. With Devin Outposts, the agent loop stays in Devin Cloud.
Your infrastructure
Yes, for teams that want background agents with local or private models, their own model keys, provider-rate billing, and the option to self-host. Both start agents from a person or a team tool and return work for review.
Yes. Devin offers models from Anthropic, OpenAI, Google, Cognition, and open-source families such as DeepSeek and Kimi. Felan also supports local and private models, and lets you bring your own keys or subscription.
When you want one vendor for the agent, a CLI, and a desktop app, and are comfortable paying for usage in Devin credits. Cognition says Devin works best on clearly specified tasks of roughly 4–8 hours of junior-engineer work.
Yes, on the Enterprise plan. Devin's Customer Dedicated Deployment is hosted by Cognition, and Outposts keep inference and planning in Devin's cloud.
Felan Team is €99 per month for the whole team, not per seat, and model usage is billed at provider rates with no markup. Devin Teams starts at $80 per month with $40 full seats, and usage is billed in credits.
Sources: Devin self-serve plans, Devin Outposts overview, Devin enterprise deployment overview, Cognition GitHub organization, Devin usage and billing, Devin models, Devin's 2025 performance review.
Start with one repo and one recurring task.
