explainx.ai350K+ learn from us
TrendingAI News TodayPathwaysSkills
Pricing
explainx.ai

Upskill in AI — 16 free pathways, live workshops & bootcamps, and 50+ courses from practitioners. Plus the skills, tools, and MCP servers to practice on.

follow us

follow on google

Add explainx.ai as a preferred source

corporate training

support@explainx.ai

get started

Find your pathTake Free Evaluation

community

Join the community

learn

mind: share how you thinkpathways — start freeworkshopsbootcampscoursescompare Explainxcertificationsmock testsexplainx universitycorporate traininglearn skills & mcp

discover

skillsmcp serversexplainx mcpestacktoolsmdx readeragentsllmsdesignsdictionarypeopleagi trackerfelony benchranks

company

aboutvisionmissionteaminstructorsteach on explainxpartnershipscommunityhackathonscareers

content

daily AI newsstate of AI — live resultsblogreleasespromptsgeneratorsresource libraryfor LLMsexplainx.ai kids

solutions

all solutionsdeveloper upskillingmarketing upskillingproduct manager upskillingleadership upskilling

newsletter · weekly

Get AI news, tools, and insights in your inbox.

supportcontactprivacytermsdata rightshow we create contentsubmission guidelines

© 2026 AISOLO Technologies Pvt Ltd

explainx.ai

On this page

  • TL;DR — what people are asking
  • Why "nearly half a year of silence" is the actual news
  • The three things MiMo scaled
  • The technical pushback worth taking seriously
  • Open-sourcing "piece by piece"
  • Why "grader compute" is its own scaling axis, not an afterthought
  • What "livestreaming a training run" actually reveals — and doesn't
  • Honest limitations
  • What this means for builders and researchers
  • Related on explainx.ai
← Back to blog

explainx / blog

Xiaomi MiMo-V2.6: Livestreaming a Trillion-Parameter RL Training Run

Xiaomi, MiMo, Reinforcement Learning, Open Source AI, Chinese AI Labs

Part of AI Research

Xiaomi's MiMo team is livestreaming the RL training run of MiMo-V2.6, scaling compute, multi-harness agentic environments, and grader compute after nearly six months of silence.

Sep 17, 2026·9 min read·Yash Thakker
add explainx.ai
go deep
Xiaomi MiMo-V2.6: Livestreaming a Trillion-Parameter RL Training Run

September 17, 2026 — Fuli Luo, who leads Xiaomi's MiMo AI team (and previously worked at DeepSeek), announced that MiMo-V2.6 is in the middle of a large-scale reinforcement learning (RL) training run — and that the team is livestreaming it publicly. The announcement follows what Luo described as "nearly half a year of silence," spent studying one specific question: how far can RL actually scale?

TL;DR — what people are asking

table · 2 cols
QuestionAnswer
What's happening?MiMo-V2.6's RL training run, livestreamed in progress
Where can I watch?mimo.xiaomi.com/rl/
What three things did they scale?Compute, environments/harnesses, grader compute
How much compute per step?~2 billion tokens per step, 1,568 prompts × 16 rollouts
Is it fully async?Team claims yes; one reply questions whether the math actually supports "fully" async
Will methodology be open-sourced?Yes — "piece by piece over the coming weeks"
What's the underlying thesis?"RL is one of the most scalable and efficient paths toward self-improvement"
Notable reaction?An engineer running a similar setup at home (122B model, 16 rollouts, 200k context) replied directly to the thread
Daily briefFrom Yash · 350K+ students taught

Catch up on AI

One AI story explained in plain English, plus one thing to try. Every day, free. Unsubscribe anytime.

Why "nearly half a year of silence" is the actual news

Luo's framing is unusual for a model announcement — most launches lead with capability claims, not a description of what the team spent six months not shipping. "We spent it studying one problem: how far RL can scale," she wrote, positioning MiMo-V2.6 less as a product launch and more as the output of a sustained research bet on reinforcement learning as the primary lever for further model improvement, rather than continuing to scale pretraining compute alone.

That framing matters because it's an implicit claim about where the field's marginal gains are coming from in late 2026: not bigger pretraining runs, but bigger, better-designed RL runs on top of an existing base model. It's the same directional bet explainx.ai has tracked across Xiaomi's earlier robotics foundation-model work and the broader industry shift toward agentic RL as the dominant post-training paradigm.

The three things MiMo scaled

Luo's thread names three specific scaling axes, each with a concrete number attached — a level of specificity that's genuinely useful for anyone trying to replicate or benchmark against this work:

table · 3 cols
AxisWhat was scaledSpecifics given
ComputeTraining-step scale~2 billion tokens per step; 1,568 prompts × 16 rollouts each; run fully asynchronously
Environments & harnessesTask diversityMulti-task agentic RL, mixed across multiple harnesses within a single training run
Grader computeReward-signal qualityAgentic in-group credit assignment, combining test-case rewards and rubric-based rewards

The "environments and harnesses" axis is worth dwelling on. Mixing multiple agentic harnesses inside one RL run — rather than training separately per-harness and merging afterward — implies the team is trying to build a single policy that generalizes across different tool-use and task structures simultaneously, rather than a narrower specialist model. That's a materially harder training-engineering problem than single-harness RL, since reward signals, episode lengths, and action spaces can differ substantially between harnesses.

The technical pushback worth taking seriously

Not every reply was celebratory. Several raised specific, substantive engineering questions rather than general skepticism:

  • Yifan Jiang challenged the "fully async" claim directly: "trainer time + sampler time = step time, so it's not 'fully async'" — pointing out that if the reported step time is simply the sum of trainer and sampler time, that's evidence of at least some sequential dependency between the two phases, not true overlap. The MiMo team hasn't yet published a response clarifying the actual pipeline overlap.
  • Dinghuai Zhang questioned the design choice to run "flash RL" and "pro RL" simultaneously, without an explained rationale in the thread itself.
  • One independent developer, replying as LLMWildling, reported doing comparable work independently at smaller scale: "122B running 16 rollouts at 200k context" — a useful external data point suggesting the general RL-scaling recipe (large rollout counts, long context) is being explored by more than one team concurrently, at different budget tiers.

None of these amount to a debunking — they're the kind of detailed technical questions a livestreamed, in-progress run invites precisely because it's unusually transparent. That transparency is itself the interesting part: publishing training internals live, rather than only in a post-hoc paper, opens the process to exactly this kind of real-time scrutiny.

Open-sourcing "piece by piece"

Luo committed to releasing the methodology incrementally — "we'll open-source the details piece by piece over the coming weeks" — rather than as a single comprehensive technical report. That's a different release cadence than the typical single-paper-plus-weights drop from labs like Moonshot AI or Alibaba's Qwen team, and it means anyone tracking this story should expect follow-up posts rather than one definitive writeup.

This continues a pattern of unusually open compute disclosure from Chinese AI labs in 2026 that explainx.ai has tracked closely — see the detailed parameter and training disclosures around Kimi K3's 2.8 trillion parameter open-weights release and Qwen 3.8 Max's open-weight token strategy — a contrast worth noting against the comparatively guarded training-compute disclosures typical of OpenAI, Anthropic, and Google DeepMind.

Why "grader compute" is its own scaling axis, not an afterthought

Most public discussion of RL scaling focuses on the two more intuitive axes — how much compute goes into generating rollouts, and how diverse the training environments are. The MiMo team's explicit call-out of grader compute as a third, independently scaled axis is worth dwelling on, because it's frequently the least-discussed bottleneck in agentic RL despite arguably mattering as much as the other two. A rollout is only as useful for training as the reward signal attached to it — if the grading step (deciding how good a given rollout actually was) is cheap, noisy, or coarse, no amount of rollout-generation compute compensates for a weak or miscalibrated signal.

"Agentic in-group credit assignment" combined with "test-case and rubric-based rewards" suggests the MiMo team is running its own agentic evaluation process to score each rollout relative to others in its group — effectively spending real inference compute on the grading step itself, not just on generating candidate behavior. That's a materially more expensive approach than a simple binary pass/fail reward or a lightweight heuristic score, and treating it as a first-class scaling axis (alongside rollout compute and environment diversity) implies the team's internal experiments found grading quality to be a genuine bottleneck worth spending compute against directly, rather than an implementation detail to minimize cost on.

What "livestreaming a training run" actually reveals — and doesn't

It's worth being precise about what a public training stream can and can't show an outside observer. Live metrics like loss curves, reward trends, or rollout throughput give real-time visibility into whether training is proceeding stably — useful for anyone trying to diagnose their own RL runs against a working reference. What a stream can't show is the accumulated design decisions that got the team to a working configuration in the first place: which reward formulations were tried and discarded, which environment mixes failed to converge, or what infrastructure bugs consumed weeks before the currently-streaming run began. Luo's framing of "nearly half a year of silence" spent on this exact question suggests a substantial amount of that trial-and-error already happened before the public stream started — meaning what's being livestreamed is closer to a validated configuration in its execution phase than an from-scratch, watch-everything-including-the-failures research process. That's still valuable transparency, but it's a different kind of transparency than showing the messy middle of the research itself.

Honest limitations

  • No benchmark results yet. This is a training run in progress, not a released model — MiMo-V2.6's actual capability level relative to competitors is unknown until training completes and evaluations are published.
  • The "fully async" claim has an unresolved technical objection from the community that the MiMo team hasn't yet addressed publicly.
  • "Piece by piece" open-sourcing has no committed timeline beyond "over the coming weeks" — treat this as an ongoing story, not a closed one.
  • No parameter count for MiMo-V2.6 itself is confirmed in this specific thread, distinct from the separate community report of a 1-trillion-parameter MiMo RL run circulating the same day.
  • Livestreaming a training run is a form of marketing as well as transparency — it generates sustained public attention across the training period, which is a real incentive independent of the scientific value of openness.

What this means for builders and researchers

Teams doing their own RL post-training: the specific numbers here — 1,568 prompts × 16 rollouts, ~2B tokens/step, mixed multi-harness environments — are a useful reference point for scaling your own agentic RL setup, even before MiMo-V2.6's final benchmarks land.

Anyone evaluating open vs. closed model strategy: this is another data point in the pattern of Chinese labs treating detailed training methodology disclosure as a competitive differentiator rather than a risk, worth weighing against closed-lab alternatives when picking a model family to build on.

Infrastructure engineers: the "fully async" debate in the replies is worth following even independent of MiMo-V2.6 specifically — trainer/sampler overlap is a genuine, generalizable RL-infrastructure problem, and a clear public answer from the MiMo team (if one comes) would be useful reference material regardless of which framework you use.

Related on explainx.ai

Update — September 22, 2026: MiMo-V2.6 has shipped — Flash, Pro, and a 9B distill, all open weights, with a 558-point Hacker News launch thread praising the training transparency over the raw benchmarks. See Xiaomi MiMo-V2.6 launches: open weights, radical training transparency for the final specs and reception.

  • Kimi K3 open weights: 2.8 trillion parameters
  • Qwen 3.8 Max preview: open-weight token strategy
  • Xiaomi Robotics U0: world foundation model via embodied synthesis
  • Scalable oversight: RLHF, Constitutional AI, weak-to-strong generalization
  • Top Chinese AI companies and startups guide
  • Moonshot AI detention rumor and the Anthropic distillation theory

Update — September 18, 2026: Reporting on the livestream puts the run's cost at roughly $493,000 per day — a rare, concrete data point on what training a trillion-parameter-class model with heavy RL/agentic-environment compute actually costs at this scale, worth comparing against Claude's own compute-to-safety-work breakdown for a sense of how frontier labs allocate similar budgets.

Official source: @_LuoFuli on X (September 17, 2026); live training stream at mimo.xiaomi.com/rl/

Details reflect Xiaomi MiMo's September 17, 2026 announcement of an in-progress training run — final model specifications and benchmark results were not yet published as of this writing.

Spotted something out of date? Let us know.
Daily briefFrom Yash · 350K+ students taught

Catch up on AI

One AI story explained in plain English, plus one thing to try. Every day, free. Unsubscribe anytime.

  • ✓ Free
  • ✓ One email a day, no spam
  • ✓ Unsubscribe anytime
Yash Thakker

Written by

Yash Thakker

Yash is an AI expert with over 300K learners. Join his workshops →

View Yash Thakker in People in AI →

Related posts

Nov 6, 2026

Baseten Loops: Serverless SFT and RL Training, Billed Per Token (What the Docs Confirm)

Baseten Loops is a Python SDK for training open-weight models with supervised fine-tuning and reinforcement learning. Reports say it goes generally available on November 6, 2026. We read the docs: three models train serverless and bill per token, 28 more need dedicated GPUs. Here is how it works and what is not yet confirmed.

Nov 4, 2026

Axion 1.0: An Open Agent Model Trained With Two SFT Stages and RL

Axion 1.0 claims frontier-level scientific reasoning and tool use from a recipe of two supervised stages plus reinforcement learning. We read the paper page, the model card and the repo to separate what is released from what is claimed.

Oct 15, 2026

TRL Tree Packing for AsyncGRPO: The 26x Claim vs. the 3x the Pull Request Measures

Hugging Face TRL merged tree packing for its AsyncGRPO trainer on October 13, 2026. Social posts cite 26x less trainer compute on MiMo Envs, while the pull request itself reports up to 3.1x on a microbenchmark. Here is the mechanism, the numbers, and how to try it.