MiMo-V2.6-Pro and MiMo-V2.6-Flash by @XiaomiMiMo have landed in Agent Arena! MiMo-V2.6-Pro ranks #5 among open-source models with +3.17% net improvement across 8.1K+ real-world agentic sessions. This release is a +10.4 percentage-point lift and a 9-spot improvement from MiMo-V2.5-Pro at #13 with -7.23%! Its strongest signal is Confirmed Success (explicit user feedback that the task worked) where it scored +7.35% and ranks #2 among open-source models! MiMo-V2.6-Flash landed #9 among open models with -0.57% net improvement across 13K+ real-world agentic sessions. At a $0.04 median cost per task (56% lower than MiMo-V2.6-Pro), it also landed on the Agent Arena Pareto frontier! See its placement below. Congrats to the @XiaomiMiMo team on the release!
Xiaomi MiMo
@XiaomiMiMo
Sep 21
Introducing Xiaomi MiMo-V2.6 — Pro & Flash. Frontier intelligence, all the modalities, built in public. 🔹 Two omnimodal models, advancing through scaled reinforcement learning 🔹 Pro performs on par with Claude Opus 5 and GPT-5.6 Sol across most agent benchmarks 🔹 Pro scores 46 on the Artificial Analysis Intelligence Index — the highest among open-source models 🔹 Stronger coding, computer use, 3D reasoning and creative capabilities 🔹 Open model weights, technical report, RL environments and training code Blog：mimo.xiaomi.com/mimo-v2-6

Oct 1, 2026 · 6:57 PM UTC

29
17
361
38,361
MiMo-V2.6-Flash has landed on the Agent Arena Pareto frontier with -0.57% net improvement at a $0.04 median cost per task! Dive into the Agent Arena Pareto frontier at: arena.ai/leaderboard/agent/p…
3
2
37
8,259
Sort replies: Relevant Recent Liked
Replying to @arena @XiaomiMiMo
Is arena ai down? Cause it wont let me in the site
4
1
84
Replying to @arena @XiaomiMiMo
8.1k sessions and flash still lands at -0.6. nice speed dial.
232
Replying to @arena @XiaomiMiMo
MiMo V2.6 Pro jumps to 5 open source in Agent Arena 10.4pp vs v2.52 on Confirmed Success Flash hits 9 and lands on the cost Pareto frontier at $0.04 task
161
Replying to @arena @XiaomiMiMo
$0.04 a task. Does the unit still clear?
75
Replying to @arena @XiaomiMiMo
@grok hy4 preview es multimodal?
1
233
Replying to @arena @XiaomiMiMo
MiMo V26 Pro hits #5 open source in Agent Arena 317% 10 4pts up from V25 ranking #2 for Confirmed Success while Flash hits #9 at $004 task on the Pareto frontier
102
Replying to @arena @XiaomiMiMo
I was a huge user of 2.5, and I love 2.6 flash, Xiaomi is cooking
102
Replying to @arena @XiaomiMiMo
MiMo-V2.6 looks like a meaningful jump, especially with that cost-to-performance balance.
26
Replying to @arena @XiaomiMiMo
Impressive agentic gains with efficiency across both releases
48
Replying to @arena @XiaomiMiMo
Impressive gains congrats to the team on this strong release
12
Replying to @arena @XiaomiMiMo
Huge congrats to Xiaomi MiMo on the impressive Agent Arena climb!
117
Replying to @arena @XiaomiMiMo
MiMo-V2.6-Pro and Flash from Xiaomi arrive in Agent Arena ranking fifth among open source agentic models
83
Replying to @arena @XiaomiMiMo
你们为什么吧battle模式给删了
115
Replying to @arena @XiaomiMiMo
Pro going from -7.2% to +3.2% in one release is the big move, but Flash at $0.04 median per task is the one I'd actually run.
79
Replying to @arena @XiaomiMiMo
开源里冲到第5，还能在8.1K多真实Agent会话里净提升3.17%，这比单看榜单分数靠谱多了。一次往上挪9位，这版MiMo是真下功夫了。
59
Replying to @arena @XiaomiMiMo
Huge progress for open models especially on confirmed success
15
Replying to @arena @XiaomiMiMo
Great progress for MiMo
45
Replying to @arena @XiaomiMiMo
我已经有点期待罗福丽MIMO V3的全新架构大模型了
42
Replying to @arena @XiaomiMiMo
stealthmodels.com/benchmarks… Mimo 2.6 is not holding up that well, far behind qwen 27b

Mimo V2.6 Flash | StealthMark

StealthMark SVG benchmark. Mimo V2.6 Flash: 43.6/100. Explore the artwork and compare results.

stealthmodels.com
47