# Xiaomi MiMo-V2.6: Pro, Flash and UltraSpeed fact check

- URL: https://ai-on-mac.com/articles/xiaomi-mimo-v2-6-pro-flash-ultraspeed-fact-check-en/
- Language: English (United States)
- Published: 2026-09-22
- Updated: 2026-11-03
- Category: Model News
- German version: https://ai-on-mac.com/artikel/xiaomi-mimo-v2-6-pro-flash-ultraspeed-faktencheck/
- Publisher: Julian Dominic Altmann, AI on Mac / KI auf dem Mac (https://ai-on-mac.com)

> Xiaomi MiMo-V2.6 Pro, Flash and UltraSpeed: correct API cost calculations, cache rates, 1M context and Responses API limits.

**Verdict:** Two open MoE models plus a hosted latency mode: pricing and benchmarks documented; independent reproduction is still pending for Flash and UltraSpeed.

**Updated October 8, 2026.** Xiaomi has released the MiMo-V2.6 family, publishing open weights for **MiMo-V2.6-Pro** and **MiMo-V2.6-Flash** while offering **MiMo-V2.6-Pro-UltraSpeed** as a hosted high-speed mode. MiMo-V2.6-Flash costs $0.14 per million input tokens and $0.28 per million output tokens, MiMo-V2.6-Pro costs $0.435 and $0.87, and UltraSpeed costs $4.35 and $8.70 — all sharing the same 1M context window and the same 131,072-token output limit. The API IDs are `mimo-v2.6-flash`, `mimo-v2.6-pro`, and `mimo-v2.6-pro-ultraspeed`.

  ![Schematic overview of the Xiaomi MiMo-V2.6 family: Pro, Flash and UltraSpeed with parameters and roles.](https://ai-on-mac.com/images/articles/xiaomi-mimo-v2-6/en/mimo-v26-architecture-light.svg)

## Specs and pricing

| Feature | Flash | Pro | Pro UltraSpeed |
|---|---:|---:|---:|
| Positioning | efficiency | flagship | low-latency / high-throughput Pro |
| Architecture | Sparse MoE | Sparse MoE | Pro family |
| Total parameters | 309B | 1.02T | same Pro family |
| Active parameters | 15B | 42B | same Pro family |
| Context | 1,048,576 | 1,048,576 | 1,048,576 |
| Max output via Responses API | 131,072 | 131,072 | 131,072 |
| Inputs | text, image, video, audio | text, image, video, audio | text, image, video, audio |
| Uncached input | $0.14/M | $0.435/M | $4.35/M |
| Cache hit | $0.0028/M | $0.0036/M | $0.036/M |
| Output | $0.28/M | $0.87/M | $8.70/M |
| Batch API | $0.07/M in, $0.14/M out | $0.2175/M in, $0.435/M out | not listed |

The open-source section of Xiaomi's launch post names Pro and Flash weights, while UltraSpeed is introduced under Desktop/API access. That makes it more accurate to treat UltraSpeed as an inference mode rather than a third independently released weight checkpoint.

## Pro vs Flash: how large is the capability gap?

Xiaomi's release card reports Pro ahead on most listed evaluations, but the gap varies substantially:

| Benchmark | Pro | Flash | Pro − Flash |
|---|---:|---:|---:|
| DeepSWE v1.1 | 71.9 | 67.9 | +4.0 |
| ProgramBench | 26.5 | 26.0 | +0.5 |
| AutomationBench v1.0.6 | 53.1 | 52.3 | +0.8 |
| Toolathlon-Verified | 76.9 | 73.6 | +3.3 |
| Agents' Last Exam | 31.6 | 27.6 | +4.0 |
| Terminal Bench 4.0 | 34.9 | 28.8 | +6.1 |
| OSWorld-Verified | 82.0 | 80.8 | +1.2 |
| CyberGym | 94.0 | **95.1** | **−1.1** |
| SEC Bench Pro | 66.3 | 47.5 | +18.8 |
| MiMo VisualCoding | 72.3 | 71.5 | +0.8 |

These are **vendor-reported results**, not a uniform independent test suite. They are still useful for understanding Xiaomi's intended positioning, but they should not be converted into a single universal "percent better" figure.

Benchmark quality also matters. DeepSWE v1.1 contains 113 long-horizon software-engineering tasks, and Epoch AI's audit identified issues in at least 23 of them. Terminal-Bench and CyberGym likewise emphasize that agent configuration and run variance can materially affect scores.

  ![Benchmark comparison Xiaomi MiMo-V2.6 Pro vs Flash according to the Xiaomi model card.](https://ai-on-mac.com/images/articles/xiaomi-mimo-v2-6/en/mimo-v26-benchmarks-light.svg)

## The independent datapoint that matters most right now

Artificial Analysis reports **46** on its Intelligence Index for MiMo-V2.6-Pro and roughly **134.3 output tokens/s** on Xiaomi's API, with pricing matching Xiaomi's published Pro rate.

That gives Pro a useful independent cross-check shortly after launch. Comparable independent runs for Flash and UltraSpeed were not yet present in the reviewed source set, so Xiaomi's up-to-20× UltraSpeed claim should be read as a vendor maximum rather than guaranteed real-world acceleration.

## MiMo API costs: correct cache-hit arithmetic

Xiaomi's [official overseas pricing](https://mimo.mi.com/docs/en-US/price/pay-as-you-go) continues to list the following **USD pay-as-you-go rates per million tokens**. Cache-hit inputs replace regular input tokens; they do **not** get charged twice.

| Official overseas API | Flash | Pro | UltraSpeed |
|---|---:|---:|---:|
| Input, cache miss | **$0.14** | **$0.435** | **$4.35** |
| Input, cache hit | **$0.0028** | **$0.0036** | **$0.036** |
| Output | **$0.28** | **$0.87** | **$8.70** |
| Batch input / batch output | $0.07 / $0.14 | $0.2175 / $0.435 | not offered |

**Worked usage scenarios, recalculated October 8, 2026.** All token counts are in millions, and cached input is *a subset of total input*, not an extra line item:

| Scenario | Total input | Cache hits | Output | Flash | Pro | UltraSpeed |
|---|---:|---:|---:|---:|---:|---:|
| Light | 5M | 0 | 1M | **$0.98** | **$3.045** | **$30.45** |
| Mid-size agent | 50M | 35M | 10M | **$4.998** | **$15.351** | **$153.51** |
| Heavy agent | 500M | 400M | 100M | **$43.12** | **$131.94** | **$1,319.40** |

**Formula:** `(total_input − cached_input) × uncached_rate + cached_input × cache_hit_rate + output × output_rate`. For Pro with 50M input, 35M cache hits and 10M output: `15 × $0.435 + 35 × $0.0036 + 10 × $0.87 = $15.351`. The former article's **$19.35 Pro** and **$193.51 UltraSpeed** figures were arithmetic errors. Actual cache reuse depends on the request and platform behavior.

  ![MiMo-V2.6 official USD uncached input, cache-hit and output rates for Flash, Pro and UltraSpeed, including corrected agent billing examples.](https://ai-on-mac.com/images/articles/xiaomi-mimo-v2-6/en/mimo-v26-price-light.svg)

For **U.S. customers**, these are API list prices; applicable sales taxes, gateway markups and paid tools may be additional. Xiaomi lists **$5 per 1,000 overseas web-search/parsing uses**. UltraSpeed is precisely 10× Pro on ordinary input and output, but Xiaomi **does not offer UltraSpeed through Batch API**.

## When UltraSpeed can make sense

UltraSpeed is primarily a latency/throughput purchase. Xiaomi says it can deliver up to 20× inference speed, while OpenRouter frames it as the same Pro checkpoint served through a much faster path.

That can be rational for interactive coding, low-latency agents or any workflow where wall-clock delay is expensive. For asynchronous research, background jobs or batch processing, the 10× token premium is difficult to justify unless measured time savings produce a larger operational gain.

## Why Flash deserves its own evaluation

Flash is not a tiny model. It is a 309B-total, 15B-active MoE with the same 1M context and the same four input modalities listed for Pro. On several Xiaomi-reported evaluations it trails Pro by only one to four points, while API pricing is dramatically lower.

For high-volume agent workloads, the relevant metric is not benchmark score per dollar in isolation. Measure **successful task cost**: token spend plus retries, recovery steps and human correction. Flash can win economically even when Pro is more accurate per attempt.

## Why Pro still matters

Pro scales to 1.02T total parameters with 42B active per token. Xiaomi's table shows its biggest margin over Flash on some long-horizon security tasks, while general-agent and visual results are often much closer.

If one failed long-running task triggers expensive reruns, Pro's higher per-token price may be offset by fewer failures. That is a workload hypothesis, not something the public benchmark table alone proves.

## Responses API compatibility: what coding agents must handle

Xiaomi's [Responses specification](https://mimo.mi.com/docs/en-US/api/chat/responses) uses `https://api.xiaomimimo.com/v1/responses`. All three V2.6 IDs support up to **131,072 output tokens**, counting both final-answer and reasoning tokens. However, it is **not feature-identical to OpenAI's Responses API**:

- `background`, `previous_response_id` and `context_management` are **unsupported**, so relying on server-side continuation by response ID is unsafe.
- Non-`auto` values of `tool_choice` are currently stripped or treated as `auto`. Forced tool execution cannot be assumed.
- `reasoning.effort=none` turns thinking off. All other accepted values currently enable the same effort behavior; `max` is **not** a special higher-compute tier.
- With reasoning enabled, sampling values are forcibly set to `temperature=1.0` and `top_p=0.95`.
- Multi-turn agents should replay prior reasoning content with tool-call history; Xiaomi's OpenCode notes warn about HTTP 400 when required `reasoning_content` is omitted.
- **JSON-output syntax differs by endpoint:** Chat Completions documents `response_format: {"type":"json_object"}`; the Responses interface documents `text.format`. Do not blindly reuse the same body between endpoints.

  ![Xiaomi MiMo-V2.6 Responses API compatibility limits for history continuation, tool_choice and reasoning effort, plus Coding Token Plan restrictions.](https://ai-on-mac.com/images/articles/xiaomi-mimo-v2-6/en/mimo-v26-routes-light.svg)

### The coding Token Plan is not a general-purpose API subscription

Xiaomi's [Token Plan terms](https://mimo.mi.com/docs/en-US/price/token-plan) permit credits in **coding tools** such as OpenCode and Claude Code but explicitly prohibit obvious non-coding traffic from arbitrary scripts or application backends. **$6/month Lite does not buy an unlimited API backend.** Its monthly credit pool is metered and depends on the selected model. Xiaomi's [newer FAQ](https://mimo.mi.com/docs/en-US/quick-start/faq/token-plan/Plans%26Pricing) includes V2.6 Pro and Flash, whereas some older pricing tables still mention V2.5. Check the live plan dashboard for the currently eligible model IDs and quota conversion.

**V2.5 is retired:** Xiaomi shut off the `mimo-v2.5` and `mimo-v2.5-pro` model IDs on **October 21, 2026 at 10:00 a.m. Beijing time**; since October 14, requests had already been routed to V2.6 Flash and V2.6 Pro ([TechNode](https://technode.com/2026/10/21/xiaomi-retires-mimo-v2-5-models-as-api-service-shifts-to-multimodal-replacements/), [LiteLLM](https://docs.litellm.ai/blog/mimo_v2_6)). Use V2.6 names and regression-test tool history and prompt behavior.

## Open weights does not mean easy local deployment

Both Pro and Flash are published on Hugging Face with MIT license metadata. But Xiaomi's reference SGLang examples are cluster-scale: Pro uses configurations including tensor parallel 16 and two nodes, while Flash is shown with tensor parallel 8.

For Mac users, the defensible conclusion today is therefore narrower: the weights are available, but Xiaomi's reference serving setup is designed for substantial GPU infrastructure. Quantized Mac-local performance needs a separate, reproducible test rather than an assumption based on the word "open."

## What was unusual about the training run

Xiaomi publicly streamed roughly six days of reinforcement-learning training. The company says Pro and Flash each completed 30 RL steps, costing about **$2.62M for Pro** and **$0.85M for Flash**, with roughly 750,000 trajectories in total.

The model cards describe a mixed RL run spanning coding, general agents, visual tasks and cybersecurity, with 1,568 prompts × 16 rollouts per update and billions of tokens per step. TechNode and NYU Shanghai RITS documented the live dashboard before release, while warning that the displayed benchmark checkpoints were Xiaomi-run rather than independently submitted leaderboard results.

## MiMo-V2.6: two open-weight models and a premium speed mode

MiMo-V2.6 is best understood as **two open-weight models plus a premium speed mode**. Flash minimizes API cost, Pro raises capability headroom, and UltraSpeed trades a 10× token price for much faster Pro inference.

The useful production test is therefore not "which row has the highest benchmark score?" It is cost per successful task, retry rate, agent-step count and overall pipeline latency on your own workload.

For a price-performance comparison, see [Qwen3.8-Flash](https://ai-on-mac.com/articles/qwen3-8-flash-en/) and [DeepSeek V4.1 Flash](https://ai-on-mac.com/articles/deepseek-v4-1-flash-en/).

## FAQ

**What is the difference between MiMo-V2.6-Pro, Flash and UltraSpeed?**

Flash is the efficiency-oriented MoE with 309B total and 15B active parameters, Pro is the flagship with 1.02T total and 42B active, and UltraSpeed is a hosted Pro-family high-throughput inference mode, not a separate weight release.

**Do Flash and Pro both have 1M context?**

Yes. Xiaomi's official model cards list 1,048,576 tokens for both, and the Responses API exposes all three V2.6 IDs with up to 131,072 output tokens.

**Can I use the $6 Lite Token Plan as an unlimited API backend?**

No. The Token Plan is metered and restricted to coding tools. Xiaomi prohibits clearly non-coding custom backends and arbitrary scripts, and standard API usage follows separate pay-as-you-go prices.

**Which API tier is cheapest?**

Flash, at $0.14/M uncached input and $0.28/M output on Xiaomi's overseas pay-as-you-go pricing. Pro costs about 3.11× Flash on uncached input, and UltraSpeed is exactly 10× Pro.

**Are the models open source?**

The weights are publicly available on Hugging Face and the model cards carry MIT license metadata. The precise phrasing supported here is 'open weights under MIT license'. Mac-local tokens/sec has not been officially measured for V2.6.

**Is UltraSpeed a stronger checkpoint than Pro?**

No based on the reviewed sources. Xiaomi and OpenRouter both describe UltraSpeed as a faster Pro-family inference mode, not a separately rated checkpoint. The price premium is a latency/throughput purchase.

## Sources

1. https://mimo.mi.com/docs/en-US/quick-start/faq/token-plan/Plans%26Pricing
2. https://mimo.mi.com/docs/en-US/quick-start/usage-guide/text-generation/structured-output
3. https://mimo.mi.com/docs/zh-CN/news/latest/v2-6
4. https://mimo.mi.com/docs/en-US/price/pay-as-you-go
5. https://mimo.mi.com/docs/en-US/api/chat/responses
6. https://mimo.mi.com/docs/en-US/api/chat/openai-api
7. https://mimo.mi.com/docs/en-US/quick-start/summary/first-api-call
8. https://mimo.mi.com/docs/en-US/api/guidance/rate-limit
9. https://mimo.mi.com/docs/en-US/quick-start/faq/api-integration
10. https://mimo.mi.com/docs/en-US/price/token-plan
11. https://mimo.mi.com/docs/en-US/tokenplan/integration/codex-configuration
12. https://mimo.mi.com/docs/en-US/tokenplan/integration/opencode
13. https://mimo.mi.com/docs/en-US/tokenplan/integration/mimo-desktop
14. https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL
15. https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL
16. https://huggingface.co/collections/XiaomiMiMo/mimo-v26
17. https://github.com/XiaomiMiMo/MiMo
18. https://artificialanalysis.ai/models/mimo-v2-6-pro
19. https://artificialanalysis.ai/models/mimo-v2-6-pro/providers
20. https://openrouter.ai/xiaomi/mimo-v2.6-pro
21. https://openrouter.ai/xiaomi/mimo-v2.6-flash
22. https://openrouter.ai/xiaomi/mimo-v2.6-pro-ultraspeed
23. https://technode.com/2026/09/18/xiaomi-livestreams-mimo-v2-6-reinforcement-learning-runs/
24. https://rits.shanghai.nyu.edu/ai/xiaomi-mimo-v2-6-live-rl-dashboard/
25. https://runtimewire.com/article/xiaomi-open-sources-mimo-v2-6-rl-cost-3-47m
26. https://deepswe.datacurve.ai/
27. https://epoch.ai/benchmarks/deepswe/review
28. https://programbench.com/
29. https://www.tbench.ai/
30. https://www.cybergym.io/cybergym/
31. https://github.com/SEC-bench/SEC-bench-Pro
32. https://github.com/MaxIntelligenceAgency/ALE-Benchmark
33. https://www.designarena.ai/about
34. https://www.reddit.com/r/LocalLLaMA/comments/1wmnw89/xiaomimimomimov26flashrl_hugging_face/
35. https://opencode.ai/de/data/xiaomi/mimo-v2-6-pro