Loading page
Xiaomi
MiMo-V2.6-Pro
MiMo-V2.6-Pro is a chat-oriented large language model from Xiaomi, part of the company's MiMo model line, with measured output of roughly 121 tokens per second.
Input from
$0.435 / 1M tokens
across 1 provider
API Pricing
| Provider | Input / 1M | Output / 1M | Cached / 1M |
|---|---|---|---|
| $0.435 | $0.870 | $0.0036 |
Prices updated daily. Last check: Sep 22, 2026
MiMo-V2.6-Pro pricing by provider
Input, output, and batch rates, plus alternatives, for one provider at a time.
Performance & Benchmarks
Source: Artificial Analysis →Intelligence
46.3 / 100
Output Speed
121 t/s
Latency (TTFT)
1.2s
Reasoning & Knowledge
- Humanity's Last Exam49.4%
Coding
- SciCode60.9%
Instruction & Long Context
- Long-Context Reasoning86.3%
Benchmarks measured Sep 2026. Scores are independent evaluations, not vendor-reported.
Model Details
General
- Creator
- Xiaomi
- Modalities
- Text
Capabilities
- Tool Calling
- No
- Open Source
- No
Strengths & Limitations
Strengths
- Measured output throughput of approximately 121 tokens per second in Artificial Analysis testing, suitable for streaming responses to interactive users
- Time to first token around 1.2 seconds, keeping the perceived startup delay in the low single-digit second range
- "Pro" tier within the MiMo V2.6 line, positioning it above lighter variants in the same generation
- Chat-tuned for instruction following and multi-turn dialogue rather than raw text completion
- Adds a non-US, non-frontier-lab option to vendor diversification strategies for teams that want alternatives to the most widely deployed model families
- Independent third-party performance measurement available, rather than creator-reported numbers only
Limitations
- We do not currently track a confirmed context window for MiMo-V2.6-Pro, so long-document workloads need verification against provider documentation
- Published benchmark results on standard reasoning, coding, and knowledge suites are not tracked in our database for this model
- Multimodal support (image or audio input) is not something we have confirmed; assume text-in/text-out until a provider states otherwise
- Smaller provider ecosystem than established model families, which can mean fewer regions, fewer SDK integrations, and less community tooling
- Ecosystem documentation and English-language evaluation coverage are thinner than for models from larger labs
Key Features
•Chat/instruction-tuned model for multi-turn conversation
•Approximately 120.97 output tokens per second measured throughput
•Approximately 1,228 ms time to first token
•Pro tier of the MiMo V2.6 generation
•Streaming token output for interactive interfaces
•Developed by Xiaomi as part of its MiMo language model series
•Third-party performance profile published via Artificial Analysis
About MiMo-V2.6-Pro
MiMo-V2.6-Pro is a conversational large language model released under Xiaomi's MiMo naming, positioned as the "Pro" variant within that version line. Xiaomi is a comparatively recent entrant among organizations publishing general-purpose language models, and the MiMo series is the company's line of assistant-style models intended for instruction following and multi-turn dialogue rather than single-shot completion.
On the serving side, independent measurements collected by Artificial Analysis put MiMo-V2.6-Pro at approximately 120.97 output tokens per second with a time to first token of about 1,228 milliseconds. Those two numbers describe different aspects of the experience: the throughput figure governs how quickly a long answer streams out once generation starts, while the roughly 1.2-second latency to the first token is what a user perceives as the initial pause after hitting send. Both are provider- and load-dependent, so figures on any given endpoint may differ from the benchmark run.
Beyond those serving metrics, computeprices.com does not currently track a confirmed context window, modality list, or public benchmark suite results for MiMo-V2.6-Pro, so readers evaluating it for a specific workload should check the documentation of whichever provider they intend to use. In practice, a model in this performance band is most often compared against other mid-to-upper tier chat models on a cost-per-token and latency basis, which is what the pricing table on this page is for.
Common Use Cases
MiMo-V2.6-Pro fits general assistant workloads where response streaming speed matters: customer-facing chat, in-product help and Q&A, drafting and rewriting text, summarization, and conversational front-ends where a roughly 1.2-second initial delay followed by ~121 tokens per second of output feels responsive. As the Pro variant of its generation, it targets tasks that need more capability than a lightweight chat model while still running at interactive speeds. Teams building agent loops or batch pipelines should first confirm the context window and any tool-calling support with their chosen provider, since those specifics are not tracked here. It is also a reasonable candidate for teams doing multi-vendor evaluations who want to price a Xiaomi-built option alongside more common model families.
Frequently Asked Questions
How much does MiMo-V2.6-Pro cost to run?
Pricing depends on the provider hosting the model and on the pricing type — per-million input tokens, per-million output tokens, and any cached-input or batch rates are all billed separately and change frequently. Check the pricing table on this page for current rates from each tracked provider rather than relying on a figure quoted elsewhere.
What is MiMo-V2.6-Pro best used for?
It is a chat model, so it suits conversational assistants, drafting and editing, summarization, and Q&A interfaces. Its measured throughput of about 121 output tokens per second and roughly 1.2-second time to first token make it a fit for interactive, user-facing applications where responses stream as they generate.
Who makes MiMo-V2.6-Pro?
Xiaomi. MiMo is the company's series of large language models, and V2.6-Pro is the Pro tier within the V2.6 generation of that series.
How fast is MiMo-V2.6-Pro?
Artificial Analysis measured approximately 120.97 output tokens per second with a time to first token of about 1,228 milliseconds. Actual figures vary by provider, region, prompt length, and server load, so treat these as a reference point rather than a guarantee.
Does MiMo-V2.6-Pro accept image input?
We do not have confirmed multimodal support for this model in our database. If your workload requires image or audio input, verify against the documentation of the specific provider you plan to use before committing.
How does MiMo-V2.6-Pro compare to other chat models?
Without tracked benchmark suite results, the most concrete comparison points we have are serving metrics — roughly 121 output tokens per second and about 1.2 seconds to first token — plus whatever per-token price the providers in the table above charge. For quality comparisons, run your own evaluation on representative prompts from your workload.