Xiaomi debuts open-weight omnimodal models MiMo-V2.6 Pro and Flash; Pro allegedly performs “on par with Opus 5 and GPT-5.6 Sol across most agent benchmarks”
Frontier intelligence, all the modalities, built in public. Today, we are releasing and open-sourcing the MiMo-V2.6 series.
Xiaomi
Context & Ripple Effects
Xiaomi has been building the MiMo line in public: it introduced MiMo-V2 in March and followed with an MIT-licensed MiMo-V2.5 release in April, positioning those models around agentic tasks. MiMo-V2.6 extends that progression from text-centric agent models to an omnimodal open-weight series.
The release is also being framed as more than a weights drop. Public discussion highlights accompanying RL task environments, an end-to-end framework, and training details; that makes Xiaomi's earlier open-weight MiMo-V2-Flash strategy a broader reproducibility play. Xiaomi's parity comparison with Opus 5 and GPT-5.6 Sol remains an allegation rather than an independently established result.
First-order effects
Developers can evaluate and deploy Xiaomi's open-weight MiMo-V2.6 Pro and Flash for multimodal work without relying solely on closed models from the vendors Xiaomi names in its benchmark comparison.
Xiaomi makes its agent-model program more useful to researchers and builders by pairing the release with reported RL tooling and task environments, rather than publishing model weights alone.
Second-order effects
Teams assessing Opus 5 or GPT-5.6 Sol gain an additional open-weight option to test against proprietary systems, even though Xiaomi's claimed performance parity requires independent validation.
Publishing RL machinery shifts some differentiation from the finished model toward the quality of training environments and implementation know-how available around it.
Third-order effects
If model makers increasingly release weights alongside reproducible training tooling, open ecosystems can compete not only on inference access but on how quickly outside developers can adapt agent systems.
The split between freely available model components and differentiated hosted products is likely to sharpen, making portability a more consequential factor in enterprise model selection.
The trend: Frontier-model competition is broadening from proprietary endpoint performance to open-weight, multimodal stacks that developers can inspect, adapt, and run across their own infrastructure.
“We're open-sourcing Pro and Flash, MiMo-V2.6-Distill-Qwen-9B, the technical report, 7K+ RL task environments, an end-to-end RL framework and composable mini-harnesses.” wow 🔥
The difference between the frontier of closed source and open source has been reduced by a good margin AND this time it is a “not” a TOP-3 Chinese labs.
Xiaomi is contributing a LOT to open research with this release. All of their training run details are basically out there in the open — reward plots, hyperparameters, data mixtures, even costs. They also release a subset of their RL envs and a smaller model for people to play ar…
How?!! Xiaomi has just released their new OPEN SOURCE model MiMo-V2.6-Pro: - multimodal text/image/audio/video - score ~ same as GPT-5.6-Sol Max - 9x cheaper input tokens - 23x cheaper output tokens 💀 Also about 20x cheaper than Opus 5... and the weights are already on Hugging Fa…
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 most insane part, they will release ~7k RL training data and the framework leading to this top 6 model on AA, they also shipped the model + tech report less than 1 week after starting the final RL run pushing both intelligence and openness level, huge congrats
Alert: Xiaomi just open-sourced MiMo-V2.6. Two natively omnimodal models, Pro and Flash, under MIT license 🔥. Pro scores 46.32 on the Artificial Analysis Intelligence Index, the highest of any open model to date. weights: https://huggingface.co/...
MiMo-V2.6-Pro debuts as the top open weights model on the Artificial Analysis Intelligence Index (46). At $0.13 per Intelligence Index task, it lands on the Intelligence vs. Cost per Task Pareto frontier @Xiaomi has just released MiMo-V2.6-Pro, an open weights model with major ad…