Which API should you use with nanoMuse?

For nanoMuse, the open-source Muse-style personal agent from Zhejiang University, the best API once the account's starting allowance is used up is a cost-optimized OpenAI-compatible endpoint. Standard Compute is in nanoMuse's provider catalogue from its next release (PR #482, merged October 16, 2026): in the web console open Connections → Chat model, click Standard Compute, paste your key and Save. On 1.0.5 and earlier, click Add an endpoint under Local or your own endpoint instead, enter https://api.stdcmpt.com/v1 as the Base URL with your key, click Add endpoint, then pick the chat model standardcompute and Save. Either way, Standard Compute's router sends each step to the cheapest model that can do it (Claude Fable 5.1 or GPT-6 Astra only when it counts), on a plan from $19 a month with a $19 compute budget. nanoMuse's own docs point to OpenRouter outside mainland China and Alibaba Cloud Bailian inside it, both billed per token, or a ChatGPT plan sign-in. nanoMuse is free, non-profit and GPL-3.0.

Guide updated 2026-10-15. Check each provider's current terms below.
Get your API key$0.25 trial for eligible new accounts, no card. Plans from $19/mo.
Light or occasional use
Start on the allowance a signed-in nanoMuse account gets on nanoMuse Cloud. After that, a DeepSeek or OpenRouter key on an efficient model will likely cost the least for light chat, and a local model through Ollama costs nothing per token.
Daily use, cost-optimized
When nanoMuse works through goals, routines and its morning feed every day, a Standard Compute endpoint sends each step to the cheapest model that can do it, and for every model we vet providers for price, uptime and speed, with automatic fallback. Opt-in Europe or US Compliance rules apply per key, team or company, and we don't train on your prompts or store them. The bill stays at the monthly plan price: requests stop with HTTP 402 when the compute budget is used, with nothing billed on top.

API and subscription options for nanoMuse

OptionPriceWhat you get
OpenRouterPay-as-you-goChoose models and providers through one API. Usage-based pricing can suit light or bursty work; estimate long agent sessions and review current fees and limits. Source: openrouter.ai
DeepSeek APIPay-as-you-goDeepSeek's own models billed per token with a DeepSeek platform key, the default route in DeepSeek Harness, Codewhale and Reasonix. Usually the cheapest route if DeepSeek models are all you use; check the current per-token and cache-hit prices. Source: api-docs.deepseek.com
Standard Compute (us)From $19/moSmart routing sends each request to the cheapest model that can do it, or pin any model by its exact ID. Standard Compute starts at $19/month with a $19 monthly compute budget. Requests stop when that budget is exhausted until renewal or a plan change; there are no automatic overage charges. Requests are not slowed as the budget is used. It is a fixed bill with a finite allowance.
ChatGPT plan (Codex sign-in)$20 Plus, $100–$500 ProSign in with ChatGPT and the agent runs on Codex models within your plan's Codex allowance: five-hour windows on Plus, none on Pro, and weekly limits may apply. Only for agents that support ChatGPT sign-in. Source: learn.chatgpt.com

Prices and allowances can change. nanoMuse's provider list, the custom endpoint entry, its config.toml [llm] block and the four model rows come from its own-key, configuration and app docs and its provider catalogue, and the add-endpoint labels from its web console's Connections screen and the desktop's own-key panel, read on October 15, 2026. The built-in Standard Compute entry comes from PR #482, merged October 16, 2026 and listed under Unreleased in nanoMuse's changelog; the latest release on October 16 was 1.0.5. nanoMuse PR #482: Standard Compute in the provider catalogue · nanoMuse: bring your own key · nanoMuse: configuration · nanoMuse repo. Spot an error? Tell us.

Pick Standard Compute in nanoMuse, or point a runtime at it

config.toml
[llm]
provider = "openai"
model    = "standardcompute"
base_url = "https://api.stdcmpt.com/v1"
api_key  = "${STANDARDCOMPUTE_API_KEY}"

From nanoMuse's next release, Standard Compute is a built-in provider with the address and the default chat model standardcompute filled in: in the web console, Connections → Chat model, click Standard Compute, paste your key and Save; on the desktop, paste the key on the Standard Compute row under Ways on in Settings → nanoMuse Cloud; on the phone, Settings → Providers → Add Provider, pick Standard Compute and paste the key. On 1.0.5 and earlier, add it as your own endpoint: in the web console, Connections → Chat model → Local or your own endpoint → Add an endpoint, with Name (optional), Base URL, API key (stored in the runtime's vault) and Models (optional) left empty so the list comes from /models, then Add endpoint and Save (desktop: Settings → nanoMuse Cloud → Any OpenAI-compatible endpoint → Add an endpoint). A runtime of your own reads the [llm] block above. The built-in entry covers chat only; the screen-operating row needs a model that sees images, so pin one there (for example anthropic/claude-sonnet-5.5) or leave it on Automatic.

nanoMuse questions

What happens when the nanoMuse Cloud allowance runs out?
nanoMuse's docs give two ways on that need nothing from nanoMuse: a key of your own at a model provider, or a plan you already pay for (ChatGPT, Claude, Kimi) signed in from the app. Your account, invite code and devices stay; only the model provider changes, and once your own key or plan is in use nothing you say passes through nanoMuse Cloud.
Is Standard Compute a built-in provider in nanoMuse?
Yes, from nanoMuse's next release. PR #482, merged October 16, 2026, adds Standard Compute to the provider catalogue the desktop, phones, relay and web console share, with base URL https://api.stdcmpt.com/v1 and the default chat model standardcompute; the changelog lists it under Unreleased, and the latest release on October 16 was 1.0.5. Then you pick Standard Compute and paste your key: Connections → Chat model in the web console, the Standard Compute row under Ways on in the desktop's Settings → nanoMuse Cloud, or Settings → Providers → Add Provider on the phone. The catalogue entry covers chat; on 1.0.5 and earlier, add Standard Compute as your own OpenAI-compatible endpoint.
Can nanoMuse use a custom OpenAI-compatible gateway?
Yes. In the web console, Connections → Chat model lists your own endpoints under Local or your own endpoint, with Add an endpoint last: enter the Base URL (usually ending in /v1) and the API key, and leave Models (optional) empty when the endpoint lists its models at /models. On the desktop the same list is under Settings → nanoMuse Cloud → Any OpenAI-compatible endpoint. A custom endpoint covers chat; the hands, picture and clip models count once their ids are set. A base URL saved from the app without a path gains /v1. On a runtime of your own, set [llm] provider = "openai" with base_url, model and api_key in config.toml.
Why can't my custom endpoint operate the screen in nanoMuse?
The screen-operating row (the hands) sends screenshots, so it needs a model that sees images, and a custom endpoint only covers the jobs you set for it. Keep chat on standardcompute and pick a vision-capable model for the hands, such as anthropic/claude-sonnet-5.5 from the same endpoint, or leave that row on Automatic.
Can I use Standard Compute with nanoMuse from mainland China?
Claude, GPT and other models are subject to the regional policies of Anthropic, OpenAI and the other model providers, and mainland China is outside their supported regions. Standard Compute is for developers in regions those providers support, for example Chinese-speaking users in Singapore or Malaysia. Inside mainland China, nanoMuse's docs recommend Alibaba Cloud Bailian first.

Try one task before choosing a plan

  1. Check the API format and model configuration supported by your installed nanoMuse version.
  2. Run a small task on a clean branch and review the result, elapsed time and compute used.
  3. Estimate your actual monthly workload before selecting a larger allowance.
Get your API keynanoMuse setup guide →Compare compute budgets →

Record your own results with the workload CSV and report template. These are blank templates, not published benchmark results.

Disclosure: this guide is maintained by Standard Compute, one of the options above. Choose based on task quality, supported features and total cost for your workload. Light usage may cost less on a usage-based API; sustained workloads need enough allowance on any subscription.

All flat-rate plans compared →The cheapest-tokens math →Plan compute for your team →