Which API should you use with MiniMax Code?

For MiniMax Code (mcode), MiniMax's open-source terminal coding agent, the best API depends on the models you want. For MiniMax models only, MiniMax's own M Plan or pay-as-you-go API costs less than any router. For Claude, GPT, GLM or Qwen next to MiniMax, Standard Compute is listed in MiniMax Code's provider picker: run /model, choose + Add 3rd-party provider…, pick Standard Compute under Known providers and paste your key. Its 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 $20 compute budget. Bringing your own key needs no MiniMax login.

Light or occasional use

If MiniMax-M3 does your work, stay with MiniMax: the API lists M3 at $0.30 input / $1.20 output per million tokens, and the M Plan starts at $22 a month. A router adds nothing for MiniMax-only use.

Daily use, cost-optimized

Standard Compute fits when MiniMax Code should reach other vendors' models too: the router 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 is 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 MiniMax Code

Prices and allowances can change. MiniMax Code's provider picker, its models.dev source and the China mirror come from its source at v0.6.5 (provider-presets.service.ts, provider-presets.client.ts and the TUI onboarding), and the BYOK command from its README and examples, read on October 16, 2026. MiniMax prices are from MiniMax's pricing and M Plan pages, read the same day. MiniMax Code repo · MiniMax Code: choose your own model · MiniMax Code: provider presets source · MiniMax pricing. Spot an error? Tell us.

Pick Standard Compute under Known providers, or add it from the shell

# In the TUI: /model → + Add 3rd-party provider… → type "standard"
# → Known providers: Standard Compute → Tab, paste key → Enter on standardcompute, Enter

# Or from the shell (the key stays in an environment variable):
export STANDARDCOMPUTE_API_KEY=YOUR_API_KEY
mcode provider add --name standardcompute --base-url https://api.stdcmpt.com/v1 \
  --api-format openai-completions --model standardcompute \
  --api-key-env STANDARDCOMPUTE_API_KEY --use

The Known providers entry comes from models.dev, which lists Standard Compute with the base URL https://api.stdcmpt.com/v1 and one model, standardcompute (the router). Review the Base URL on the model screen (Ctrl+E edits it); MiniMax Code tests the model before it saves anything. To pin other models such as anthropic/claude-sonnet-5.5, choose Custom provider instead and use Import models from /models, or repeat --model in the shell command.

MiniMax Code questions

Is Standard Compute built into MiniMax Code?

It is in the provider picker. MiniMax Code fills its Known providers list from models.dev, and models.dev lists Standard Compute (base URL https://api.stdcmpt.com/v1, model standardcompute), so in MiniMax Code 0.6.5 you run /model, choose + Add 3rd-party provider…, type standard and pick Standard Compute. Paste your key, then MiniMax Code tests the model and saves it as a custom provider. MiniMax's own providers are not in that list; you reach them with mcode login.

Do I need a MiniMax account to use my own API key in MiniMax Code?

No. MiniMax Code's README says bringing your own key does not require a MiniMax login: add the provider with /model or mcode provider add and run mcode. A MiniMax sign-in (mcode login) is only for MiniMax's own models and plans.

Is MiniMax's own plan cheaper than Standard Compute for MiniMax Code?

For MiniMax models only, yes. MiniMax lists MiniMax-M3 at $0.30 input / $1.20 output per million tokens, and its M Plan starts at $22 a month; Standard Compute serves the same M3 at the same list price, so a router adds nothing there. Standard Compute is worth it when you also want Claude, GPT, GLM or Qwen on one key with each step sent to the cheapest model that can do it, from $19 a month with a $20 compute budget.

Why don't I see Standard Compute in MiniMax Code's provider list in China?

In the China region, MiniMax Code does not read models.dev directly: it reads a copy MiniMax mirrors on filecdn.minimax.chat, pinned to a release, so a provider can appear there later than on models.dev. If Standard Compute is missing, choose Custom provider and enter https://api.stdcmpt.com/v1, OpenAI Compatible and your key. Note that Claude, GPT and other models follow the regional policies of Anthropic, OpenAI and the other model providers, and mainland China is outside their supported regions.