Skip to main content
AI Choice Engine

Xiaomi

Xiaomi MiMo-V2.6-Flash

Xiaomi MiMo-V2.6-Flash is a Xiaomi model with 1.05M context, $0.14 in / $0.28 out per million tokens, last verified October 25, 2026. Open weights: yes. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.

MiMo-V2.6-Flash is Xiaomi's low-cost open-weight omni-modal MoE with a 1M context at $0.14/$0.28.

Recorded profile verification October 25, 2026Individual provider fields may change

AI model specification details
SpecificationXiaomi MiMo-V2.6-Flash
ProviderXiaomi
TierBalanced
Context window1.05M
Max output131K
Input / 1M tokens$0.14
Output / 1M tokens$0.28
WeightsOpen
Parameters309B total / 15B active (MoE)
Reasoning levelsNot verifiedUnverified
Modalitiestext, image, video, audio
LicenseMIT
API model idmimo-v2.6-flash
ReleasedSeptember 22, 2026

Pricing tiers: Xiaomi overseas realtime USD/MTok input/output $0.14/$0.28, cached input $0.0028; Batch $0.07/$0.14, cached $0.0014. Domestic RMB/MTok realtime input/output ¥1/¥2, cache ¥0.02; Batch ¥0.50/¥1, cache ¥0.01. Cache writes remain limited-time free with no end date. Overseas search is $5/1,000 calls; domestic ¥16/1,000. Token Plan quotas are separate from PAYG balance. MIT checkpoint discloses 309B total/15B active parameters.

Publisher evidence

Published benchmark results

Each result keeps its source date visible. Dates refer to source publication or capture unless the source explicitly states when it measured the result. A missing benchmark is not treated as a zero.

Published benchmark results for Xiaomi MiMo-V2.6-Flash
BenchmarkScoreSource dateSource
Artificial Analysis Intelligence IndexMiMo-V2.6-Flash382026-10-01Artificial Analysis · View source

Best for

  • Cheap omni-modal input
  • Open-weight deployments
  • High-volume long-context work

Watch out

Thinking defaults enabled via thinking.type=enabled/disabled; no graded effort enum is documented. The 131,072 completion budget includes reasoning and final output. Thinking forces temperature=1.0/top_p=0.95. Complete reasoning_content must pass through subsequent tool-call turns or the API can return 400. Native checkpoint context is 1,048,576; hosted docs advertise 1M. Limits are per account/model, not per API key. The retired V2.5 alias previously routed here with a changed default completion budget; this does not merge the old and new checkpoint identities or move historical benchmark results.

Common questions

Xiaomi MiMo-V2.6-Flash

Answered from the verified figures on this page rather than general guidance.

How much does Xiaomi MiMo-V2.6-Flash cost per million tokens?
Xiaomi MiMo-V2.6-Flash lists base rates of $0.14 per million input tokens and $0.28 per million output tokens. Estimate both input and output token volume and any additional charges for your workload. Xiaomi overseas realtime USD/MTok input/output $0.14/$0.28, cached input $0.0028; Batch $0.07/$0.14, cached $0.0014. Domestic RMB/MTok realtime input/output ¥1/¥2, cache ¥0.02; Batch ¥0.50/¥1, cache ¥0.01. Cache writes remain limited-time free with no end date. Overseas search is $5/1,000 calls; domestic ¥16/1,000. Token Plan quotas are separate from PAYG balance. MIT checkpoint discloses 309B total/15B active parameters.
What is Xiaomi MiMo-V2.6-Flash's context window?
Xiaomi MiMo-V2.6-Flash accepts about 1.05M tokens of context. That only matters if you routinely send very long documents, large codebases, or multi-turn histories that approach that limit.
What is Xiaomi MiMo-V2.6-Flash best for?
Xiaomi MiMo-V2.6-Flash is a balanced tier from Xiaomi. It suits cheap omni-modal input, open-weight deployments, high-volume long-context work. Thinking defaults enabled via thinking.type=enabled/disabled; no graded effort enum is documented. The 131,072 completion budget includes reasoning and final output. Thinking forces temperature=1.0/top_p=0.95. Complete reasoning_content must pass through subsequent tool-call turns or the API can return 400. Native checkpoint context is 1,048,576; hosted docs advertise 1M. Limits are per account/model, not per API key. The retired V2.5 alias previously routed here with a changed default completion budget; this does not merge the old and new checkpoint identities or move historical benchmark results.
Can I self-host Xiaomi MiMo-V2.6-Flash?
Xiaomi MiMo-V2.6-Flash publishes open weights, but self-hosting depends on the licence, hardware footprint, quantisation quality, and serving stack. A hosted API is often cheaper until you have measured throughput and concurrency on your own hardware.