DaoXEOne API and its popular fork New API are open-source panels for relaying and managing LLM API keys: you self-host the panel, add upstream channels (each with its own source key/base URL), and it exposes a unified OpenAI-compatible endpoint with tokens, quotas and multi-user billing. Crucially, the panel is not itself a model source — it's only as honest and reliable as the channels you plug in, which is exactly where quiet model-swapping can happen. DaoXE, by contrast, is the managed access itself. Verify current panel features in their repos.
Why people compare them#
- Unified OpenAI endpoint. Both end up as one OpenAI-compatible URL for many models, so they get compared.
- Panel vs access. One API/New API is a management/reselling layer you host; DaoXE is the model access underneath.
- Where swapping hides. A relay panel is only as trustworthy as its upstream channels — so verification matters even more. Learn to detect model swapping.
Head-to-head#
| Aspect | DaoXE | One API / New API |
|---|---|---|
| What it is | Managed model access | Self-hosted relay/key-management panel |
| Hosting | Nothing to host | You deploy the panel (Docker) & keep it updated |
| Upstream channels | Not applicable — DaoXE is the source | You source & configure each upstream channel |
| Model-swap risk | Vendor invites benchmark verification | As trustworthy as the channels you plug in — detect model swapping |
| Native Anthropic Messages | Yes — /v1/messages | Depends on channel/version support |
| Best for | Direct managed access, no ops | Teams reselling/managing keys across upstreams |
| Region | Overseas-ready; not for mainland China | Runs wherever you host it |
| Cost model | Usage-based — live pricing | Free software; you pay/operate the upstreams |
Which one fits you#
- Pick One API / New API if you want to self-host a panel to manage or resell keys across upstream channels you already have.
- Pick DaoXE if you want actual managed access without hosting a panel or sourcing upstreams — from a vendor that invites benchmark verification.
- Whatever you relay through, verify. With any panel, quietly-swapped models are the risk; diff at temperature 0 and detect model swapping.
Try it in minutes#
No panel to deploy or channels to source — point your OpenAI SDK at DaoXE and call a model:
from openai import OpenAI
# DaoXE: nothing to host; one key already spans GPT / Claude / Gemini / DeepSeek...
client = OpenAI(base_url="https://daoxe.com/v1", api_key="YOUR_DAOXE_KEY")Verify us — don't trust us
Point the open benchmark at DaoXE and at the official API and compare at temperature 0. Then learn to detect model swapping so a cheaper endpoint can't quietly swap you to a smaller model.
Frequently asked questions#
Are One API / New API model providers?
No — they're self-hosted relay/management panels. You still supply the upstream channels. DaoXE is the managed access itself.
Do these panels prevent model swapping?
They don't guarantee it — a relay is only as honest as its upstream channels. Always benchmark and detect model swapping, whatever the panel.
Do I have to host anything with DaoXE?
No — DaoXE is managed. One API/New API require you to deploy and maintain the panel plus its upstreams.
Can I put DaoXE behind One API / New API?
You can add an OpenAI-compatible upstream channel — but if you just want access, calling DaoXE directly is simpler.
Try DaoXE — and benchmark it yourself
One key for GPT, Claude, Gemini, DeepSeek and more. Point the open benchmark at us and compare — don't take our word for it.