MiMo API Token Cost Calculator

As of July 2026, MiMo-V2.5-Pro list pricing is $0.42 per 1M input + $0.83 per 1M output, with cache-hit input at $0.0035 per 1M (99% off from the 2025 launch rate). MiMo-V2.5 standard is $0.14 input / $0.28 output list, with cache-hit input at $0.0028 per 1M. V2-Flash was deprecated on 2026-06-30 along with the rest of the V2 series. MiMo-7B is free under the MIT open-weight license. See the full pricing hub →

The cache-hit rate is the percentage of input tokens that benefit from prompt caching. When you send a request, the API server can match a prefix of your prompt against previously-cached tokens (system prompts, retrieved documents, conversation history) and bill those tokens at the much lower cache-hit rate. Production estimates:

• Single-shot summarization / RAG (no repeated prompt): 0–20%
• Multi-turn chat: 50–80% (most input is prior conversation)
• Agent with persistent system prompt + tool docs: 70–95%
• Code review with same file corpus each time: 80–95%

Yes. MiMo-V2.5-Pro list input is ~97% cheaper than Claude Opus 4.5 ($0.42 vs $15.00 per 1M) and ~99% cheaper on cache-hit input ($0.0035 per 1M). For coding agents and long-context workloads the cost gap is even larger because MiMo's 1M context window removes the need to chunk documents.

Three official channels: (1) Xiaomi MiMo Platform, (2) OpenRouter, (3) HuggingFace Pro. All three expose OpenAI-compatible endpoints.

For MiMo, one token is approximately 0.75 English words or 1.5 Chinese characters (BPE tokenizer). The calculator assumes a 3:1 input-to-output ratio is a reasonable starting estimate for chat workloads; replace it with your actual token count for production planning.

Prices are sourced from official provider pages and reflect list pricing as of July 20, 2026. Volume discounts, batch API discounts, and provider-specific caching nuances are NOT included. For mission-critical budgeting, always confirm with the provider's pricing page.