Xiaomi MiMo API Pricing 2026: MiMo-V2.6-Pro Costs
MiMo-V2.6-Pro UltraSpeed costs $4.35 per million input tokens and $8.70 output on Xiaomi's API. Here's what every MiMo model costs, what UltraSpeed buys you, and how to run it for half price, sealed if you need it.

On this page
Prices checked 19 November 2026. MiMo-V2.6-Pro UltraSpeed costs $4.35 per million input tokens and $8.70 output on Xiaomi's API, as of 19 November 2026. SayGM serves the same model for $2.24 and $4.48, 48.5% less, with no subscription and no fee on buying credit. The sealed version, which runs inside a hardware-attested enclave, costs the same $2.24 and $4.48.
MiMo API pricing, checked 19 November 2026 (per million tokens)
| Model | Where | Input per M | Output per M | Cache read per M | Checked |
|---|---|---|---|---|---|
| MiMo-V2.6-Pro UltraSpeed | SayGM | $2.24 | $4.48 | $0.019 | 19 Nov 2026 |
| MiMo-V2.6-Pro UltraSpeed, sealed | SayGM | $2.24 | $4.48 | $0.019 | 19 Nov 2026 |
| MiMo-V2.6-Pro UltraSpeed | Xiaomi | $4.35 | $8.70 | $0.036 | 19 Nov 2026 |
| MiMo-V2.6-Pro UltraSpeed | OpenRouter (incl. 5.5% fee) | $4.59 | $9.18 | $0.038 | 19 Nov 2026 |
| MiMo-V2.6-Pro UltraSpeed | Requesty (incl. 5% markup) | $4.57 | $9.14 | $0.038 | 19 Nov 2026 |
| MiMo-V2.6-Pro | Xiaomi | $0.435 | $0.87 | $0.0036 | 19 Nov 2026 |
| MiMo-V2.6-Pro, batch | Xiaomi | $0.2175 | $0.435 | $0.0018 | 19 Nov 2026 |
| MiMo-V2.6-Flash | Xiaomi | $0.14 | $0.28 | $0.0028 | 19 Nov 2026 |
Overseas list prices in US dollars, before tax. Xiaomi's are from its pay-as-you-go pricing page, SayGM's from its live API, and OpenRouter's and Requesty's from their model lists plus their published fees. SayGM serves the UltraSpeed version only.
Table of Contents
- What is Xiaomi MiMo?
- What MiMo costs in a real month
- MiMo vs other open models on price
- Private MiMo: the confidential version
- Who serves MiMo: SayGM vs Xiaomi vs OpenRouter vs Requesty
- How can you pay less for the MiMo API?
- Using MiMo on SayGM
- MiMo API pricing FAQ
- TL;DR
- About SayGM
What is Xiaomi MiMo?
MiMo is Xiaomi's family of large language models. Xiaomi released and open-sourced the MiMo-V2.6 series in September 2026, in two sizes: Pro and Flash. Both have a 1M-token context window and up to 128K tokens of output, and the weights are on Hugging Face under the MIT licence.
Xiaomi's own pitch for V2.6-Pro is agent work. It says the model is on par with Claude Opus 5 and GPT-5.6 Sol on most agent benchmarks, and recommends it for "complex projects, long-term tasks, high-value work". Flash is for frequent calls and large-scale office tasks.
UltraSpeed is a high-speed mode of V2.6-Pro, which Xiaomi says delivers "up to 20x inference speed". Xiaomi recommends it for "production scenarios with strong real-time interaction and sensitivity to response speed": voice agents, live coding assistants, chat where a user is waiting. That speed is priced in. On Xiaomi's own API, UltraSpeed costs ten times standard Pro, and it has no batch option.
So the choice comes first. If your traffic can wait a few seconds, standard Pro direct from Xiaomi at $0.435 and $0.87 is the cheapest MiMo. If you need UltraSpeed, SayGM is where it costs least.
What MiMo costs in a real month

Three common workloads at list prices on 19 November 2026. Cached tokens are billed as cache reads; Xiaomi doesn't charge for cache writes at the moment.
| Workload | Xiaomi UltraSpeed | SayGM UltraSpeed (standard or sealed) | Xiaomi standard Pro |
|---|---|---|---|
| Support chatbot: 50M input, 10M output, half cached | $196.65 | $101.27 | $19.66 |
| RAG app: 200M input, 20M output, 75% cached | $396.90 | $204.40 | $39.69 |
| Coding agent: 500M input, 50M output, 90% cached | $668.70 | $344.38 | $66.87 |
On UltraSpeed, SayGM saves about $95 a month on the chatbot and about $324 on the agent. The cache matters less here than on most models: MiMo's cache reads are already under 1% of the input price, so output is most of the bill once caching is on.
The standard Pro column shows what the speed costs. For a background agent nobody is watching, standard Pro is a tenth of the price, and Xiaomi's batch tier halves it again.
MiMo vs other open models on price
Here's what 100 million tokens cost at an 80/20 input-to-output mix on SayGM, on 19 November 2026.
| Model on SayGM | Input / output per M | Per 100M tokens | Sealed version per 100M |
|---|---|---|---|
| MiMo-V2.6-Pro UltraSpeed | $2.24 / $4.48 | $269 | $269 |
| Kimi K3 | $1.93 / $9.66 | $348 | $456 |
| GLM 5.3 | $0.97 / $3.04 | $138 | - |
| Qwen3.8 27B | $0.062 / $0.37 | $12.36 | - |
| DeepSeek V4.1 Flash | $0.037 / $0.15 | $5.93 | - |
UltraSpeed sits at the top end of open models on price, but below Kimi K3, and well below Kimi's sealed version. For cheaper open models across the board, see our cheapest LLM API comparison.
Private MiMo: the confidential version
SayGM's confidential MiMo runs the open weights inside a trusted execution environment. In the model page's words: "Inference happens sealed, so the prompt stays inside the enclave." Nobody in the chain can read the prompt, SayGM included, and the hardware's attestation lets you verify what ran.
The standard version is different. It's served by a provider reached through SayGM, which receives the prompt under its API terms.
On 19 November 2026, the sealed version costs the same as the standard one: $2.24 input, $4.48 output and $0.019 for cache reads. That isn't true of every model. Kimi K3's sealed version costs about 31% more than its standard one on SayGM. Our guide to private inference covers how the sealed tier works.
Who serves MiMo: SayGM vs Xiaomi vs OpenRouter vs Requesty
Prices below are for MiMo-V2.6-Pro UltraSpeed per million input and output tokens, checked 19 November 2026.
SayGM

$2.24 and $4.48, 48.5% below Xiaomi's list, and the sealed version, the only one we found, at the same price. There's no subscription and no fee on funding your balance, and the same key serves Claude, GPT, Gemini and other open models.
Missing: standard Pro, Flash, batch, and image, audio or video input.
Xiaomi

$4.35 and $8.70. Xiaomi is the only place to get every MiMo model and mode: standard Pro at $0.435 and $0.87, Flash at $0.14 and $0.28, half-price batch for both, full multimodal input, and a $0.72 bonus on your first top-up.
Missing: a discount on UltraSpeed, which has no batch option, and published rate limits for it; Xiaomi asks UltraSpeed users to contact it for custom capacity.
OpenRouter

$4.59 and $9.18, Xiaomi's list plus OpenRouter's 5.5% fee on credit. Xiaomi is the only UltraSpeed provider OpenRouter lists, while standard Pro comes from several hosts, including DeepInfra at $0.387 and $0.783.
Missing: a price advantage on UltraSpeed.
Requesty

$4.57 and $9.14 with its 5% markup on model cost. Requesty lists UltraSpeed from Xiaomi, and standard Pro from Xiaomi, DeepInfra and Tencent.
Missing: a price advantage on UltraSpeed.
How can you pay less for the MiMo API?
- Pick the right mode. UltraSpeed is ten times the price of standard Pro on Xiaomi. Use it where response speed is the product.
- Cache long prompts. Cache reads are $0.036 per million on Xiaomi UltraSpeed and $0.019 on SayGM, under 1% of the input price, and Xiaomi doesn't charge for cache writes at the moment.
- Batch background work on standard Pro. Xiaomi's batch tier halves standard Pro and Flash. It isn't available for UltraSpeed.
- Run UltraSpeed through SayGM. It's 48.5% below Xiaomi's list on input, output and cache reads, with the sealed version at the same price. There's no subscription and no fee on funding your balance.
Using MiMo on SayGM
SayGM serves MiMo through the OpenAI-compatible chat completions API, so the OpenAI SDK works with a new base URL and key:
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.saygm.com/v1",
api_key=os.environ["SAYGM_API_KEY"],
)
response = client.chat.completions.create(
model="mimo-v2.6-pro-ultraspeed-tee",
messages=[{"role": "user", "content": "Summarise this incident report in three bullet points."}],
)
print(response.choices[0].message.content)Use mimo-v2.6-pro-ultraspeed for the standard version. Both support tool calling. On SayGM they take text only, though Xiaomi's own API also accepts images, audio and video. The live rates are on MiMo on SayGM and the confidential model page.
MiMo API pricing FAQ
How much does Xiaomi MiMo cost?
As of 19 November 2026, Xiaomi charges $0.435 per million input tokens and $0.87 output for MiMo-V2.6-Pro, $0.14 and $0.28 for MiMo-V2.6-Flash, and $4.35 and $8.70 for MiMo-V2.6-Pro UltraSpeed, at its overseas rates. SayGM serves UltraSpeed for $2.24 and $4.48.
Is the MiMo API free?
No. As of 19 November 2026, Xiaomi bills every request, but overseas users get a one-off $0.72 bonus on their first top-up, valid for 40 days. The V2.6 weights are free to download under the MIT licence if you host them yourself.
What is MiMo-V2.6-Pro UltraSpeed?
MiMo-V2.6-Pro UltraSpeed is a high-speed mode of Xiaomi's MiMo-V2.6-Pro that Xiaomi says runs up to 20 times faster, for real-time apps. It costs $4.35 per million input tokens and $8.70 output on Xiaomi as of 19 November 2026, ten times standard Pro, and $2.24 and $4.48 on SayGM.
Is MiMo open source?
Yes. Xiaomi released the MiMo-V2.6 Pro and Flash weights on Hugging Face in September 2026 under the MIT licence, which allows commercial use.
Can I use MiMo outside China?
Yes. Xiaomi runs a separate overseas platform as of November 2026: you sign up with an email, Google or Facebook account, pay in US dollars, and get your own base URL and key. You can also call MiMo-V2.6-Pro UltraSpeed through SayGM, OpenRouter or Requesty.
Is there a private or confidential version of MiMo?
Yes. SayGM runs MiMo-V2.6-Pro UltraSpeed inside a hardware-attested enclave, so nobody in the chain can read the prompt. As of 19 November 2026 it costs the same as the standard version, $2.24 per million input tokens and $4.48 output.
Where is MiMo cheapest?
It depends on the mode. As of 19 November 2026, the cheapest MiMo-V2.6-Pro is the standard version direct from Xiaomi, at $0.435 and $0.87. For UltraSpeed, SayGM is cheapest at $2.24 and $4.48, against $4.35 and $8.70 on Xiaomi and more on OpenRouter and Requesty.
TL;DR
Xiaomi MiMo API pricing for MiMo-V2.6-Pro UltraSpeed is $4.35 per million input tokens and $8.70 output on Xiaomi, as of 19 November 2026. SayGM serves it for $2.24 and $4.48, and the sealed version costs the same. If you don't need UltraSpeed, standard Pro direct from Xiaomi is a tenth of the price.
Point your OpenAI client at api.saygm.com/v1, set the model to mimo-v2.6-pro-ultraspeed-tee, and say gm to AI.
About SayGM
SayGM is a drop-in inference gateway for teams who don't want to just take a company's word that their prompts are private. Route to a confidential model and your prompt is sealed inside hardware-verified execution, where not even SayGM can see it. That's not a policy, it's provable. Swap in your existing OpenAI, Anthropic, or Gemini code and you're covered in minutes, at transparent, published rates with no hidden markup.
Say gm to AI at saygm.com.
- Pricing


