Bumo Router vs OpenRouter: Model Coverage, Fees and Payment (2026)
Bumo Router and OpenRouter both sell one API key for many models, but they are built for different buyers. Bumo Router is a focused gateway for official Claude, GPT and Gemini APIs, billed in USD at the listed model price plus a 6% fee. OpenRouter is a marketplace with a much wider catalogue, including open-weight and Chinese models, and it charges a fee when you buy credits. Everything below comes from the two vendors' live model endpoints and pricing pages, checked on October 21, 2026. I did not run a latency benchmark, so this article makes no speed claims.
The short answer
| If you need... | Start with... |
|---|---|
| The widest catalogue, including open-weight and Chinese models | OpenRouter |
| Claude, GPT and Gemini with USD card billing and one simple fee | Bumo Router |
| To pay with Alipay or USDC | OpenRouter |
| OpenAI, Anthropic and Responses endpoints from one key | Bumo Router |
| The lowest price on one specific model | Compare both live price pages; neither vendor is always cheaper |
The rest of this article explains the two gaps that matter most in practice: model coverage and how each vendor charges.
Model coverage: 27 versus 468
On October 21, 2026, Bumo Router's international /v1/models endpoint returned 27 model IDs. They cover Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5, several older Claude models, GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna, GPT-5.6, GPT-5.5, GPT-5.4, GPT-5.3 Codex, Gemini 3.8 Flash, Gemini 3.1 Pro Preview and two image models.
OpenRouter's public models endpoint returned 468 entries on the same day, including free and batch variants. The exact count changes often, but the shape does not: OpenRouter carries a much broader catalogue, including many open-weight models and models from Chinese providers.
That difference is not cosmetic. If the model you want is not on Bumo Router's list, a lower fee does not help you.
What about DeepSeek, Kimi, GLM and Qwen?
This is the question most coding-plugin users ask, so it deserves a direct answer.
Bumo Router runs two regional gateways:
- International:
router.bumo.ai, billed in USD. - China:
router.bumo.cc, billed in CNY through WeChat Pay.
The China gateway carries DeepSeek (deepseek/deepseek-v4-pro, deepseek/deepseek-flash), Kimi (kimi/kimi-k3), GLM (glm/glm-5.3) and Qwen (qwen/qwen3.8-max), among others. Those model IDs are not available on the international gateway today. The two gateways also have separate catalogues, balances and payment methods.
So if you are outside mainland China and your main goal is to run DeepSeek, Kimi, GLM or Qwen inside Cline, Kilo Code or Continue, the practical options today are OpenRouter, the model vendor's own API, or another gateway that carries them. Do not buy a Bumo Router international key expecting those models; check /v1/models first. If you mainly use Claude, GPT or Gemini, the international catalogue is the relevant one.
This is a factual limitation, not a temporary marketing omission. The gateway's model list is the source of truth, and it can change independently of this article.
Pricing: 6% on usage versus 5.5% on credits
Both vendors say they pass through the model provider's list price. The difference is where the fee sits.
Bumo Router: the international gateway charges the listed model price plus a 6% service fee, in USD. You prepay a balance (minimum $2), and the balance is valid for one year. There is no separate credit-purchase fee.
OpenRouter: model usage is passed through at provider list price. OpenRouter charges a fee when you purchase credits: 5.5% for card payments (minimum $0.80) and 5% for crypto. There is also a BYOK fee of 5% above a $25,000 per month list-price allowance.
At the same list price, Bumo Router's effective fee is about half a percentage point higher than OpenRouter's card fee. Rough arithmetic: $100 of model usage costs $106 on Bumo Router, versus roughly $105.50 on OpenRouter. For small teams that difference is minor; at scale, half a point matters, and OpenRouter's BYOK option can change the comparison again.
Here are four current models with the prices each vendor returned on October 21, 2026, in USD per 1 million tokens:
| Model | Bumo Router (input / output) | OpenRouter (input / output) |
|---|---|---|
| Claude Opus 5.5 | $4.24 / $21.20 | $4.00 / $20.00 |
| Claude Fable 5.1 | $10.60 / $53.00 | $10.00 / $50.00 |
| GPT-6.1 Sol | $2.12 / $10.60 | $2.00 / $10.00 |
| Gemini 3.8 Flash | $0.795 / $3.975 | $0.825 / $4.125 |
Two caveats. First, OpenRouter's price can vary by upstream provider; the figures above are what its public models endpoint returned, not a promise for every route. Second, Bumo Router's page describes the formula as list price plus 6%, but a few older catalogue entries do not follow that formula exactly. Use the live model page for the exact price of the model you plan to run, not a marketing summary. Gemini 3.8 Flash is one case where Bumo Router's listed price is slightly below OpenRouter's.
Payment and billing
| Bumo Router (international) | OpenRouter | |
|---|---|---|
| Payment methods | Cards via Stripe | Cards, AliPay, USDC |
| Currency | USD | USD |
| Top-up minimum | $2 | Check current checkout terms |
| Balance validity | 1 year from top-up | Check current terms |
| Fee | 6% on usage | 5.5% card / 5% crypto on credit purchase |
Bumo Router's international balance is separate from its China gateway balance. If your team needs one invoice or one account across both regions, that is not how the product is structured today.
Protocol and client setup
Bumo Router exposes OpenAI Chat Completions, Anthropic Messages and OpenAI Responses endpoints. For OpenAI-compatible clients, the base URL is:
https://router.bumo.ai/v1
For Anthropic-format clients such as Claude Code, the base URL is https://router.bumo.ai (without /v1), because the client appends the Messages path itself.
Bumo Router also supports model fallback. You can pass a models array with a primary and backup models, or set a default fallback on the API key. When the primary model's upstreams fail with a retryable error such as a rate limit or a 5xx, the gateway tries the next model.
OpenRouter exposes an OpenAI-compatible API and documents its own fallback and provider-routing options. Moving a client between the two services is mostly a change of base URL, API key and model ID.
Cline, Kilo Code and Continue
All three clients work with Bumo Router's international gateway as an OpenAI-compatible provider:
| Client | Where to configure | Base URL | Model ID examples |
|---|---|---|---|
| Cline | API Provider → OpenAI Compatible | https://router.bumo.ai/v1 | claude-opus-5-5, gpt-6.1-sol |
| Kilo Code | Settings → Providers → Custom provider → OpenAI Compatible | https://router.bumo.ai/v1 | claude-opus-5-5, gemini-3.8-flash |
| Continue | ~/.continue/config.yaml | https://router.bumo.ai/v1 | claude-opus-5-5, gpt-6.1-sol |
A minimal Continue entry looks like this:
name: Bumo Router
version: 1.0.0
schema: v1
models:
- name: Claude Opus 5.5
provider: openai
model: claude-opus-5-5
apiBase: https://router.bumo.ai/v1
apiKey: YOUR_BUMO_API_KEY
roles:
- chat
Use the exact model ID from /v1/models. In particular, do not paste a China-gateway ID such as deepseek/deepseek-v4-pro into an international configuration: it is not in that catalogue.
Latency and region
I did not benchmark either service for this article, and I would not choose between them on a vendor's "fast" claim. Both route requests to upstream model providers, so latency depends on the model, the provider, your network and the time of day.
If latency matters for your product, run the same prompt on both services from the region where your users are, then record time to first token, total completion time, error rate and output quality. A single average hides the slow requests that users actually notice.
Bumo Router's two-gateway structure matters for compliance and billing, not just models: the international gateway is USD and the China gateway is CNY through WeChat Pay, with separate accounts and balances. OpenRouter runs a single global account with provider-level routing.
What to choose
- Choose Bumo Router if your stack is Claude, GPT and Gemini, you want to pay in USD by card, and you prefer a small catalogue with a flat 6% fee over a marketplace.
- Choose OpenRouter if you need breadth, open-weight or Chinese models, Alipay or USDC, or BYOK.
- If you came here for DeepSeek, Kimi, GLM or Qwen in a coding plugin and you are outside mainland China, use OpenRouter or the model vendor's API today. Bumo Router's international gateway does not carry those models. Its China gateway does, but it is a separate CNY/WeChat product.
- If cost is the deciding factor, compare the exact model's live price and the effective fee. A half-point fee difference can be smaller than a model-level price difference.
Check the live sources
Model lists and prices change. These are the pages this comparison used:
- Bumo Router international model list: router.bumo.ai/v1/models
- Bumo Router console: router.bumo.ai
- OpenRouter model list: openrouter.ai/api/v1/models
- OpenRouter pricing and fees FAQ: openrouter.ai/docs/faq
If you mainly use Claude, GPT or Gemini, create a Bumo Router key and top up from $2. The live model list is the authoritative source for IDs and prices.