Valorant has a new game mode and it is one of the most interesting additions to the game in years thanks to the removal of some abilities.
Its iX3 Long Wheelbase model also set to get Dolby’s premium surround sound audio solution.
Mistral AI, the Paris-based artificial intelligence company valued at €11.7 billion ($13.8 billion), today released Workflows in public preview — a production-grade orchestration layer designed to move enterprise AI systems out of proofs of concept and into the business processes that generate revenue.
The product, which launches as part of Mistral’s Studio platform, is the company’s clearest articulation yet of a thesis that is quietly reshaping the enterprise AI market: that the bottleneck for organizations adopting AI is no longer the model itself, but the infrastructure required to run it reliably at scale.
“What we’re seeing today is that organizations are struggling to go beyond isolated proofs of concept,” Elisa Salamanca, who leads go-to-market for Mistral’s enterprise products, told VentureBeat in an exclusive interview ahead of the launch. “The gap is operational. Workflows is the infrastructure to run AI systems reliably across business-critical processes.”
The release arrives at a pivotal moment for both Mistral and the broader AI industry. The dedicated agentic AI market has been valued at approximately $10.9 billion in 2026 and is projected to reach $199 billion by 2034. Yet despite that staggering growth trajectory, industry research points to a stark reality: over 40% of agentic AI projects will be aborted by 2027 due to high costs, unclear value, and complexity. Mistral is betting that Workflows can help its enterprise customers avoid becoming one of those statistics.
At its core, Workflows provides a structured system for defining, executing, and monitoring multi-step AI processes — from simple sequential tasks to complex, stateful operations that blend deterministic business rules with the probabilistic outputs of large language models.
Salamanca described Workflows as containing several key components. The first is a development kit that allows engineers to build orchestration logic in just a few lines of Python code. “We have also been able to expose MCP servers,” she explained, referring to the Model Context Protocol standard for connecting AI systems to external tools, “so that they can actually do this with agent authoring.”
The second — and arguably more technically significant — component is an architecture that separates orchestration from execution. “We’re decorrelating the orchestration from the execution,” Salamanca said. “Execution can happen close to the customer’s data — their critical systems — and orchestration can happen on the cloud or wherever they want to run it.” This means the data never has to leave the customer’s perimeter, a design decision with enormous implications for regulated industries where data sovereignty is non-negotiable. “Enterprises do not have to worry about us having access to the data,” she added.
The third pillar is observability. According to Mistral’s blog post announcing the release, every branch, retry, and state change within a workflow is recorded in Studio with native support for OpenTelemetry. Salamanca noted that this is not an afterthought: “You can easily see what decisions have been taken by the workflow, by the agent, and you can deep dive into where problems are happening.”
Workflows is fully customizable across models — engineers can select which model handles which step and can inject arbitrary code, allowing them to blend deterministic pipelines with agentic sections. The system also supports connectors that integrate directly with CRMs, ticketing systems, support platforms, and other enterprise tools, with built-in authentication and secrets management.
Unlike some competitors offering drag-and-drop workflow builders, Mistral has deliberately targeted developers and engineers rather than business users. “There are a couple of solutions out there that have click-and-drag, drag-and-drop solutions for workflows,” Salamanca acknowledged. “This is not the approach that we’ve been taking. We’ve been really focused towards developers and critical systems that will not scale if you’re doing these drag-and-drop workflows.”
The decision is part of a broader philosophy at Mistral: that enterprise AI systems handling mission-critical operations — cargo releases, compliance reviews, financial transactions — require the precision and version control that only code can provide. Business users are not excluded from the picture, but their role is downstream. Once engineers write a workflow in Python, it can be published to Le Chat, Mistral’s chatbot platform, so anyone in the organization can trigger it. Every step remains tracked and auditable in Studio.
Under the hood, Workflows runs on Temporal’s durable execution engine — a platform whose $5 billion valuation reflects how its durable execution capabilities, originally built for cloud workflow orchestration, have become essential infrastructure for AI agents requiring reliable, long-running, stateful processes. Temporal’s customers include OpenAI, Snap, Netflix, and JPMorgan Chase, and its technology powers orchestration at companies like Stripe and Salesforce.
Mistral extended Temporal’s core engine for AI-specific workloads by adding streaming, payload handling, multi-tenancy, and observability that the base engine does not provide out of the box. “Workflows is built on top of Temporal,” Salamanca confirmed. “We added all the AI requirements to make these AI workflows reliable. It provides out of the box durability, retries, state management. Whenever there’s a failure, it starts again wherever it stopped.” Originally spun out of Uber’s Cadence project, Temporal transparently handles retries, state persistence, and timeouts, providing durable execution across failures. In late 2025, Temporal joined the newly formed Agentic AI Foundation as a Gold Member and announced an official OpenAI Agents SDK integration. By building on this infrastructure rather than creating a proprietary alternative, Mistral inherits battle-tested reliability while focusing its own engineering efforts on the AI-specific layer that sits above it.
Mistral is not launching Workflows as a concept — the company says customers are already running the product in production, processing millions of executions daily across three primary use cases.
The first is cargo release automation in the logistics sector. Global shipping still runs on paperwork, and a single cargo release can involve customs declarations, dangerous goods classifications, safety inspections, and regulatory checks spanning multiple jurisdictions. Salamanca described the scope of the problem: “Their global shipping today runs on paperwork. They have to involve customs declaration, Dangerous Goods classification, safety inspections, regulatory checks, and Workflows is now powering that with our models and business rules inside.”
Critically, the system keeps humans in the loop at the right moments. According to Mistral’s blog, the human approval step in a workflow is a single line of code — wait_for_input() — that pauses the workflow indefinitely with no compute consumption, notifies the reviewer, and resumes exactly where it left off once approval is given. “Humans are still in the loop, but they’re in the loop at the right time,” Salamanca said. “They just get the validation — I don’t have to go into multiple tools — and the shipment gets released.”
The second production use case is document compliance checking for financial institutions, specifically Know Your Customer reviews. These reviews are manual, repetitive, and traditionally require hours of analyst time per case. Salamanca said Workflows now processes these reviews in minutes and provides outputs in an auditable manner — a requirement for meeting regulatory obligations.
The third example involves customer support in the banking sector. “You’d have millions of users actually asking to have credit cards blocked, or feedbacks on their account situation, on their credit feedbacks,” Salamanca said. With Workflows, incoming support tickets are analyzed, categorized by intent and urgency, and routed automatically. Each routing decision is visible and traceable in Studio, and when the system gets a categorization wrong, the team can correct it at the workflow level without retraining the model.
Workflows does not exist in isolation. It is the middle layer of a three-part enterprise platform that Mistral has been assembling at a rapid clip throughout 2026.
At the bottom sits Forge, the custom model training platform Mistral launched in March at Nvidia’s GTC conference. Forge allows organizations to build, customize, and continuously improve AI models using their own proprietary data. At the top sits Vibe, Mistral’s coding agent platform that provides the user-facing interaction layer — available on web, mobile, or desktop.
Salamanca connected the three explicitly: “We just released Forge. It enables you to create your own models. But the question is, how do you put these models to do valuable work for your enterprise? That’s where Workflows comes in, because this is the orchestration piece — how you blend in deterministic rules and agentic capabilities. And then if you really want to have your end users interact with these AI patterns, it’s where Vibe comes into play.”
Forge is already seeing strong traction, Salamanca said, across two distinct patterns of enterprise demand. “First, they wanted to really build completely dedicated models to solve unique problems — transformers-based architecture for time series in the financial sector, adding new types of modalities to the LLMs,” she explained. “And the second motion was about customers with really specific tasks they want to solve. Reinforcement learning really caught their attention as to how they can use Forge and Forge RL to actually have models do these tasks very well.”
This layered architecture — model customization, workflow orchestration, and end-user interfaces — positions Mistral as something more ambitious than a model provider. It is building a full-stack enterprise AI platform, a strategy that pits it directly against not just other AI labs like OpenAI and Anthropic, but also against the hyperscale cloud providers. The company’s product portfolio now ranges, as Salamanca put it, “from compute to end-user interfaces,” including data centers in Europe, document processing with its OCR model, and audio capabilities through its Voxtral models.
The Workflows launch comes as Mistral executes one of the most aggressive scaling campaigns in the history of the European technology industry. The French AI startup has increased its revenue twentyfold within a year, with co-founder and CEO Arthur Mensch putting the company’s annualized revenue run rate at over $400 million, compared to just $20 million the previous year. The Paris-based company aims to achieve recurring annual revenue of more than $1 billion by year-end.
The company’s fundraising trajectory has been equally dramatic. Mistral announced a €1.7 billion ($1.9 billion) Series C round at a €11.7 billion ($12.8 billion) valuation in September 2025. Bloomberg reported in September 2025 that the company was finalizing a €2 billion investment valuing it at €12 billion ($14 billion). ASML led the round and contributed €1.3 billion, a landmark investment that aligned chip manufacturing expertise with frontier AI development and underscored European industrial capital’s commitment to building a sovereign AI ecosystem. Mistral then secured $830 million in debt in March 2026 to buy 13,800 Nvidia chips for a new data center near Paris.
The financial picture illustrates why Workflows matters strategically. Mistral’s revenue growth is being driven primarily by enterprise adoption, with approximately 60% of revenue coming from Europe, according to CEO Mensch’s public statements. Those enterprise customers are not buying Mistral’s models for casual chatbot applications — they are deploying them in regulated, mission-critical environments where reliability and data sovereignty are table stakes. Workflows gives those customers the production infrastructure they need to actually deploy AI systems that matter.
In May 2025, Mistral released Mistral Medium 3, which was priced at $0.40 per million input tokens and $2 per million output tokens. The company said clients in financial services, energy, and healthcare had been beta testing it for customer service, workflow automation, and analyzing complex datasets. That model now becomes one of many that can be plugged into Workflows, creating a flywheel where better models drive more workflow adoption, which in turn drives more inference revenue.
Mistral’s entry into workflow orchestration arrives in an increasingly crowded field. AI orchestration platforms are quickly becoming the backbone of enterprise AI systems in 2026, and as businesses deploy multiple AI agents, tools, and LLMs, the need for unified control, oversight, and efficiency has never been greater.
Major cloud providers — Amazon with Bedrock AgentCore, Microsoft with Copilot Studio, Google with Vertex AI’s agent tools, and IBM with WatsonX — all offer some form of workflow or agent orchestration. Open-source frameworks like LangChain, LlamaIndex, and Microsoft AutoGen provide developer-level building blocks. And dedicated orchestration startups are proliferating.
Mistral’s differentiation rests on three pillars. First, vertical integration: because Workflows is native to Studio, the orchestration layer and the components it orchestrates — models, agents, connectors, observability — are built to work together, eliminating the integration tax that enterprises pay when stitching together disparate tools. Second, deployment flexibility: the split control-plane/data-plane architecture means customers in regulated industries can run execution workers in their own environments while still benefiting from managed orchestration. Third, data sovereignty: Mistral’s European roots and infrastructure investments give it a natural advantage with organizations wary of routing sensitive data through U.S.-headquartered cloud providers — a concern that has intensified amid ongoing geopolitical tensions and growing European anxiety about relying on foreign providers for over 80% of digital services and infrastructure.
Still, the challenges are real. OpenAI and Anthropic both have significantly larger model ecosystems and developer communities. The hyperscalers control the cloud infrastructure where most enterprise workloads actually run. And the enterprise sales cycles for production-grade AI deployments remain long and complex, requiring deep technical integration work that even well-funded startups can struggle to staff.
Salamanca outlined three areas of near-term development. First, Mistral plans to release a more managed version of Workflows that abstracts deployment logic for developers who don’t need granular control over worker placement. “Whenever you want to have this flexibility, you can, but if you want to be able to have this on a managed infrastructure, even if it’s running in your own VPC, this is something that we’re adding,” she said.
Second, the company intends to make Workflows accessible to business users, not just engineers. “With Vibe code, you can actually author a workflow. This can be executed at scale, and any end user, in the end, can actually do that with Workflows,” Salamanca explained. The third area is enterprise guardrails and safety controls for agentic applications — ensuring agents use the correct tools, run with appropriate permissions, and that administrators can enforce policies at scale. “Making sure that we have all these enterprise controls to be able to scale the authoring and the building of these workflows is something we’re actively working on,” she said.
The Python SDK for Workflows (v3.0) is now publicly available. Developers can try the product in Studio and access documentation and demo templates immediately. Mistral will be hosting its inaugural AI Now Summit in Paris on May 27–28, where the company is expected to provide additional details on its platform roadmap.
For three years, the AI industry has been captivated by a single question: who can build the most powerful model? Mistral’s Workflows launch suggests the company has moved on to a different question entirely — one that may prove far more consequential for the enterprises writing the checks. It’s not about which model is smartest. It’s about which one can actually show up for work.
An iPhone software update is on its way and now Apple has let slip when it will arrive.
For the first time since the launch of the Apple App Store in 2008, Apple will allow subscribers to access discounted pricing. But it’s not for everyone.
The new D1 SSD enclosure has a rugged build quality that’s designed for use in difficult environments where precious data needs to be physically protected.
Microsoft and OpenAI on Monday announced a sweeping overhaul of the partnership that has defined the commercial AI era, dismantling key pillars of exclusivity and revenue-sharing that bound the two companies together for years and replacing them with a looser, time-limited arrangement that gives both sides far more freedom to pursue rival relationships.
The amended agreement, disclosed simultaneously in blog posts from both companies, marks the most significant restructuring since Microsoft first invested $1 billion in OpenAI in 2019 — and it transforms what was once the most consequential exclusive technology alliance in a generation into something that more closely resembles a strategic but arm’s-length commercial relationship.
Under the new terms, Microsoft will no longer pay any revenue share to OpenAI when customers access OpenAI models through Azure. OpenAI, meanwhile, will continue paying a revenue share to Microsoft through 2030 — at the same 20 percent rate — but that obligation is now subject to a total cap. Microsoft retains a license to OpenAI’s intellectual property for models and products through 2032, but that license is now explicitly non-exclusive. And OpenAI, critically, can now serve all of its products to customers on any cloud provider — including Amazon Web Services and Google Cloud — ending the exclusivity that had been a cornerstone of the original deal.
“The rapid pace of innovation requires us to continue to evolve our partnership to benefit our customers and both companies,” Microsoft wrote in its blog post Monday. OpenAI echoed the framing, calling the amended agreement a move “grounded in flexibility, certainty, and a focus on delivering the benefits of AI broadly.”
The diplomatic language belies the drama that led to this moment — months of behind-the-scenes tension, competing deal announcements, public contradictions, and even the specter of litigation between two companies whose fates have been intertwined since the earliest days of the generative AI revolution.
To understand why Monday’s announcement matters so much, it helps to understand what came before it. When Microsoft poured its initial $1 billion into OpenAI in 2019, and then followed with a cumulative investment exceeding $13 billion, it secured something extraordinary: exclusive commercial access to OpenAI’s models and intellectual property. Azure became the sole cloud provider for OpenAI’s API products. Microsoft integrated OpenAI’s GPT models into everything from Bing to Office to GitHub Copilot. The arrangement was, by any measure, one of the most lopsided technology licensing deals in modern history — Microsoft got privileged access to the most capable AI models on the planet, and OpenAI got the capital and infrastructure it needed to scale.
The deal even contained an unusual provision: Microsoft’s exclusive rights would remain in force until OpenAI achieved artificial general intelligence, or AGI — a loosely defined milestone referring to AI systems that rival or exceed human intelligence across a broad range of tasks. OpenAI’s board retained the authority to declare when AGI had been reached, at which point certain commercial terms would change. It was, in effect, a philosophical tripwire embedded in a business contract.
That structure worked well enough when OpenAI was a research lab with a modest commercial footprint. But as ChatGPT exploded into the mainstream in late 2022 and OpenAI’s annualized revenue rocketed into the billions, the constraints began to chafe. OpenAI found itself locked into a single cloud ecosystem at precisely the moment when enterprises — its fastest-growing customer segment — were demanding multi-cloud flexibility. In an internal memo earlier this month, OpenAI’s revenue chief Denise Dresser put it bluntly, telling staff that the Microsoft partnership had “limited our ability to meet enterprises where they are,” according to a report from The Verge.
The proximate cause of Monday’s restructuring was not a philosophical disagreement about AI safety or corporate governance. It was a $50 billion check from Amazon. In February, OpenAI announced that Amazon would invest up to $50 billion in the company — $15 billion upfront, with another $35 billion to follow when certain unspecified conditions were met. In exchange, OpenAI agreed to expand its existing cloud agreement with AWS by $100 billion over eight years and, most controversially, committed to making AWS the exclusive third-party distribution provider for Frontier, its new enterprise agent-building platform. OpenAI also agreed to co-develop “stateful runtime technology” on AWS Bedrock, the infrastructure layer that allows AI agents to maintain memory and context over extended tasks.
The problem was that OpenAI’s existing contract with Microsoft almost certainly prohibited these arrangements. Microsoft held exclusive rights to any OpenAI product accessed through an API — a category that plainly included Frontier. On the very day OpenAI announced the Amazon deal, Microsoft issued a pointed public statement insisting that “Azure remains the exclusive cloud provider of stateless OpenAI APIs” and that “OpenAI’s first party products, including Frontier, will continue to be hosted on Azure.” The contradiction between the two announcements was stark, and it created immediate legal exposure. The Financial Times reported in March that Microsoft was actively considering legal action to enforce its contractual rights. The situation placed OpenAI in an impossible position: it had made promises to Amazon that it seemingly could not keep under the terms of its Microsoft agreement.
Monday’s deal resolves that impasse entirely. By converting Microsoft’s license from exclusive to non-exclusive and explicitly granting OpenAI the right to serve products on any cloud, the new terms retroactively validate the Amazon arrangement and eliminate the legal overhang. Amazon CEO Andy Jassy wasted no time celebrating. “We’re excited to make OpenAI’s models available directly to customers on Bedrock in the coming weeks, alongside the upcoming Stateful Runtime Environment,” he wrote on X, adding that the company would share more details at an event in San Francisco on Tuesday.
The financial mechanics of the new deal deserve careful parsing, because they reveal which side gave up what — and who came out ahead. Under the old arrangement, money flowed in both directions. When customers bought ChatGPT subscriptions or accessed OpenAI models through their own applications, OpenAI paid Microsoft a cut — reportedly 20 percent. Conversely, when enterprise customers accessed OpenAI models through Azure’s API, Microsoft paid OpenAI a share of that revenue. This bilateral structure reflected the deep integration between the two companies: Microsoft was simultaneously OpenAI’s investor, cloud provider, distribution partner, and largest customer.
The new deal makes the cash flow one-directional. Microsoft stops paying OpenAI entirely. OpenAI continues paying Microsoft its 20 percent share, but only through 2030, and now subject to a total cap whose precise dollar figure has not been disclosed. Given that OpenAI’s revenue is growing rapidly — the company was reportedly on pace to generate tens of billions annually — that cap could become material relatively quickly.
For Microsoft, the trade-off is straightforward: it sacrifices the exclusivity that made Azure the only gateway to OpenAI’s models, but it gains immediate financial relief by eliminating its outbound revenue-share payments while continuing to collect inbound payments for several more years. And it retains approximately 27 percent ownership of OpenAI’s for-profit entity, meaning it participates in the company’s growth regardless of which cloud serves the workloads. Last quarter alone, Microsoft reported $7.5 billion in revenue from its OpenAI investment in a single quarter, according to TechCrunch’s reporting. For OpenAI, the calculus is different. It accepts a continued obligation to pay Microsoft through 2030, but it gains the commercial freedom to sell everywhere — a freedom that is arguably worth far more than the revenue-share savings. Enterprise customers overwhelmingly operate in multi-cloud environments. Being locked into Azure was not just a technical constraint; it was a sales objection that OpenAI’s competitors, particularly Anthropic and Google, exploited relentlessly.
One of the more philosophically intriguing aspects of Monday’s announcement is what it does to the AGI provision that once governed the partnership. Under the original agreement, Microsoft’s exclusive commercial rights were tied to a trigger: if OpenAI’s board determined that the company had achieved AGI, certain terms — including Microsoft’s access to the most advanced models — would change. The provision was meant to ensure that a truly superintelligent system would remain under the nonprofit board’s control rather than being commercially exploited. In practice, it created perverse incentives: OpenAI had a financial reason to never declare AGI, and Microsoft had a financial reason to argue that AGI had not been reached regardless of what the technology could actually do.
The new deal sidesteps this entirely. Microsoft’s license now runs through a fixed calendar date — 2032 — “independent of OpenAI’s technology progress,” as the companies put it. The AGI trigger, a concept that once sat at the philosophical heart of the partnership, has been replaced by a spreadsheet. Andrew Curran, a close observer of OpenAI’s governance, noted on X that language defining AGI had been removed from OpenAI’s website, sharing a screenshot showing the change. The move drew sharp reactions. One commenter observed that “removing the definition = removing the accountability. whoever controls when AGI is declared controls a lot of commercial terms.”
The shift reflects a broader maturation — or perhaps disillusionment — within the AI industry regarding AGI as a meaningful commercial or governance concept. When the original deal was struck, AGI felt like a distant, almost mythical threshold. Now, with models like GPT-5.5 demonstrating increasingly general capabilities, the term has become more of a marketing slogan than a technical benchmark. Replacing it with fixed dates and dollar caps is, in some sense, an admission that the industry has moved beyond the framework that once defined this partnership.
The most immediate beneficiary of the new arrangement is the enterprise customer. For years, organizations that wanted access to OpenAI’s models had essentially one option: Azure. That constraint is now gone. Within weeks, according to Jassy, OpenAI’s models will be available on AWS Bedrock alongside the stateful runtime environment that powers long-running AI agents. Google Cloud is presumably not far behind.
This multi-cloud availability arrives at a moment when the AI infrastructure market is undergoing rapid consolidation and expansion simultaneously. Meta recently committed $48 billion to cloud providers CoreWeave and Nebius. Amazon’s investment in OpenAI, combined with its existing relationship with Anthropic — in which Amazon has invested up to $4 billion — positions AWS as a model-agnostic platform where enterprises can mix and match AI capabilities. Microsoft, meanwhile, has developed its own relationship with Anthropic, using Claude to power agentic products — a hedge against the very OpenAI dependency it spent billions creating.
The competitive dynamics are now genuinely complex. Microsoft competes with OpenAI in AI products (Copilot vs. ChatGPT), partners with OpenAI’s rival Anthropic, and remains OpenAI’s largest shareholder. OpenAI sells on Azure, AWS, and soon everywhere else, while building its own data centers. Amazon invests in both OpenAI and Anthropic. Google builds its own models while also hosting competitors on Vertex AI. Jehangeer Hasan, a technology commentator, captured the mood on X, calling the announcement a “notable shift in the cloud AI landscape” that signals “intensifying multi-cloud competition and a push toward giving developers more flexibility instead of locking them into a single ecosystem.” Chris Alexander, an engineer, offered a more candid assessment: “honestly Azure’s OpenAI endpoints are so unreliable, we mostly just hit you all directly,” adding that “it would be nice to have options in AWS or GCP for sure.”
Several open questions remain. The precise dollar amount of the revenue-share cap has not been disclosed, and it will matter enormously as OpenAI’s revenue scales. The meaning of “first on Azure” — whether it implies a meaningful exclusivity window or merely simultaneous availability — remains deliberately ambiguous. And OpenAI’s own infrastructure ambitions, including plans to build proprietary data centers, could eventually reduce its dependence on any third-party cloud, including Azure.
Microsoft’s position, while less dominant than before, is not as diminished as some early commentary suggested. It remains OpenAI’s primary cloud provider, its largest shareholder, and a licensee of its technology through the end of the decade. It has diversified its own AI strategy with investments in Anthropic, its own Phi and MAI model families, and deep integration of AI across its product portfolio. The company reported $7.5 billion in OpenAI-related revenue last quarter — a figure that demonstrates the sheer financial scale of the relationship even in its loosened form.
For OpenAI, the new agreement is a coming-of-age moment. The company that once depended on Microsoft for everything — capital, compute, distribution, and credibility — now operates as an independent force capable of striking multi-billion-dollar deals with Microsoft’s biggest rivals. Sam Altman announced the changes on X with characteristic brevity: “We have updated our partnership with Microsoft.”
Seven years ago, when Microsoft CEO Satya Nadella and Altman first shook hands on a deal to commercialize artificial intelligence, the arrangement rested on the assumption that OpenAI needed Microsoft more than Microsoft needed OpenAI. Every clause — the exclusivity, the AGI trigger, the revenue share — reflected that original imbalance. Monday’s restructuring is proof that the assumption no longer holds. The partnership that launched the generative AI revolution has survived, but the power dynamics that created it have not. In the AI industry, it turns out, the only thing that moves faster than the technology is the leverage.
A leaked Google Photos update reveals a massive redesign to the app’s Collections tab. Here are the proposed changes.
Xiaomi, the Chinese firm best known for its smartphones and electric vehicles, has lately been shipping some incredibly affordable and high-powered open source AI large language models.
The trend continued today with the release of Xiaomi MiMo-V2.5 and Xiaomi MiMo-V2.5-Pro, both available under the permissive, enterprise-friendly MIT License, making them suitable for use in production in commercial applications. Enterprises and individual/independent developers can now download either of the models (and more Xiaomi open source options) directly from Hugging Face, modify them as needed, and run them locally or on virtual private clouds as they see fit.
The most notable attribute of these models besides the open source licensing is that, according to Xiaomi’s published benchmarks, they are among the most efficient available for agentic “claw” tasks, that is, powering systems such as OpenClaw, NanoClaw and Hermes Agent, in which users can communicate with them directly over third-party messaging apps and have the agents go off and complete tasks on the human user’s behalf, such as making and publishing marketing content, running accounts, organizing email and scheduling, etc.
As Xiaomi’s ClawEval benchmark chart shows, both MiMo-V2.5 and the Pro version in particular appear near the top left of the chart, indicating high performance in completing the benchmarked claw tasks while using the fewest amount of tokens — saving the human user money, especially in a world where more and more services such as Microsoft’s GitHub Copilot are moving to usage-based billing (charging the human behind the agents for each token used rather than imposing rate limits like Anthropic or providing an “all-you-can-eat” buffet-style subscription like OpenAI).
In fact, the Pro model leads the open-source field with a 63.8% success rate, consuming only ~70K tokens per trajectory.
This is roughly 40–60% fewer tokens than those required by Anthropic Claude Opus 4.6, Google Gemini 3.1 Pro, and OpenAI GPT-5.4 to achieve comparable results.
By combining a massive 310B-parameter architecture with a highly efficient “active” footprint and a native 1-million-token context window, Xiaomi MiMo is challenging the dominance of closed-source frontier models from Google and OpenAI, especially when it comes to the latest and greatest craze in enterprise AI deployments — agentic tasks and “claws” similar to OpenClaw.
Xiaomi has released two distinct versions of the model to serve different ends of the development spectrum: MiMo-V2.5 (the “Omni” multimodal specialist) and MiMo-V2.5-Pro (the “Agent” specialist).
While the base model provides native multimodality, the MiMo-V2.5-Pro is specifically engineered for “long-horizon coherence” and complex software engineering.
On the GDPVal-AA (Elo) benchmark, the Pro model achieved a score of 1581, surpassing competitors like Kimi K2.6 and GLM 5.1.
Xiaomi researchers further released data on several high-complexity tasks performed autonomously by V2.5-Pro:
SysY Compiler in Rust: The model implemented a complete compiler from scratch—including lexer, parser, and RISC-V assembly backend—in 4.3 hours. Spanning 672 tool calls, the model achieved a perfect 233/233 score on hidden test suites, a task that typically takes a computer science major several weeks.
Full-Featured Video Editor: Over 11.5 hours and 1,868 tool calls, the model produced an 8,192-line desktop application featuring multi-track timelines and an export pipeline.
Analog EDA Optimization: In a graduate-level engineering task, the model optimized a Flipped-Voltage-Follower (FVF-LDO) regulator in the TSMC 180nm process. By iterating through an ngspice simulation loop, the model improved metrics like line regulation by 22x over its initial attempt.
These experiments highlight a “harness awareness” in V2.5-Pro, where the model actively manages its own memory and shapes its context to sustain coherence over thousands of sequential tool calls.
Over the API, Xiaomi is pricing the models at competitive rates for both domestic (Chinese) and international markets (like the U.S.). For overseas developers, the high-performance MiMo-V2.5-Pro is priced at $1.00 per million input tokens (for a cache miss) and $3.00 for output within context windows up to 256K.
For ultra-long context tasks between 256K and 1M tokens, the cost doubles to $2.00 for input and $6.00 for output, though the architecture’s caching capabilities offer significant relief, reducing input costs to as little as $0.20 to $0.40 per million tokens upon a cache hit.
Domestically, these rates are mirrored in yuan, with the Pro model starting at ¥7.00 per million input tokens for standard context and reaching ¥14.00 for the extended 1M range. Meanwhile, the base model starts at just $0.40 USD for overseas input per million tokens and $2.00 per million output, putting it among the more affordable third of leading LLMs globally (see our chart below):
|
Model |
Input |
Output |
Total Cost |
Source |
|
Grok 4.1 Fast |
$0.20 |
$0.50 |
$0.70 |
|
|
MiniMax M2.7 |
$0.30 |
$1.20 |
$1.50 |
|
|
MiMo-V2.5 Flash |
$0.10 |
$0.30 |
$0.40 |
|
|
Gemini 3 Flash |
$0.50 |
$3.00 |
$3.50 |
|
|
Kimi-K2.5 |
$0.60 |
$3.00 |
$3.60 |
|
|
MiMo-V2.5 |
$0.40 |
$2.00 |
$2.40 |
|
|
MiMo-V2-Pro (≤256K) |
$1.00 |
$3.00 |
$4.00 |
|
|
GLM-5 |
$1.00 |
$3.20 |
$4.20 |
|
|
GLM-5-Turbo |
$1.20 |
$4.00 |
$5.20 |
|
|
DeepSeek V4 Pro |
$1.74 |
$3.48 |
$5.22 |
|
|
GLM-5.1 |
$1.40 |
$4.40 |
$5.80 |
|
|
Claude Haiku 4.5 |
$1.00 |
$5.00 |
$6.00 |
|
|
Qwen3-Max |
$1.20 |
$6.00 |
$7.20 |
|
|
Gemini 3 Pro |
$2.00 |
$12.00 |
$14.00 |
|
|
GPT-5.2 |
$1.75 |
$14.00 |
$15.75 |
|
|
GPT-5.4 |
$2.50 |
$15.00 |
$17.50 |
|
|
Claude Sonnet 4.5 |
$3.00 |
$15.00 |
$18.00 |
|
|
Claude Opus 4.7 |
$5.00 |
$25.00 |
$30.00 |
|
|
GPT-5.5 |
$5.00 |
$30.00 |
$35.00 |
|
|
GPT-5.4 Pro |
$30.00 |
$180.00 |
$210.00 |
To lower the barrier for agentic development further, Xiaomi has made cache writing free of charge for a limited time across all models, alongside a total fee waiver for the entire MiMo-V2.5-TTS suite, which includes its specialized voice cloning and design features.
This pricing logic is clearly designed to accelerate the transition from simple chat applications to persistent, long-horizon agents that can operate at a fraction of the cost of legacy frontier models.
Xiaomi has also introduced an overhauled version of its subscription offerings, called the “Token Plan,” now available in four levels:
The Lite “Starter Pack” provides 720 million credits for $63.36 USD per year
Standard tier offers 2.4 billion credits for $168.96 per year
A Pro tier provides 8.4 billion credits for $528.00 per year (designed for enterprise use cases)
Max —aimed at high-intensity coding enthusiasts—delivers 19.2 billion credits for $1,056.00 per year
Beyond credit allotments, all plans include preferential API rates, a 20% discount for off-peak calls, and “Day-0” support for popular coding scaffolds like Cursor, Zed, and Claude Code.
However, both through the API and via the Token Plan, accessing the Xiaomi models from China may present barriers or additional compliance and regulatory risks to U.S.-based enterprise customers. As such, the best bet for U.S. enterprises concerned about relying on Chinese tech but wanting to take advantage of the low cost and open source models is likely setting up their own virtual private clouds or local servers, downloading the model weights, and running the models domestically.
At the heart of MiMo-V2.5 is a Sparse Mixture-of-Experts (MoE) architecture. While the model boasts a total of 310 billion parameters, only 15 billion are “active” during any given inference cycle.
Meanwhile, V2.5-Pro is 1.02 trilion-parameter Mixture-of-Experts model with 42 billion active parameters.
In either case, the design functions much like a specialized research hospital: while the facility has hundreds of doctors (parameters), only the specific specialists required for a particular case (query) are called into the room.
This massive increase in parameter volume for the Pro version provides the “neural capacity” required for the deep, multi-step reasoning found in complex software engineering and long-horizon tasks, as though even more specialists are available in an even larger hospital.
According to Xiaomi’s blog post, the regular V2.5 follows a rigorous five-stage evolution:
Text Pre-training: Building a massive language backbone on 48 trillion tokens.
Projector Warmup: Aligning in-house audio and visual encoders with the language core.
Multimodal Pre-training: Scaling across high-quality cross-modal data.
Agentic Post-training: Progressively extending the context window from 32K to 1M tokens.
RL and MOPD: Utilizing Reinforcement Learning and Multimodal Preference Optimization (MOPD) to sharpen real-world reasoning and perception.
The backbone utilizes a hybrid sliding-window attention architecture, inherited from MiMo-V2-Flash, which optimizes how the model “remembers” long-range information. This technical foundation enables MiMo-V2.5 to see, hear, and reason natively, rather than relying on external “plug-in” tools for visual or auditory processing.
Conversely, the training of MiMo-V2.5-Pro prioritizes “action space” over sensory perception. Instead of sensory alignment, the Pro model’s training focus shifts toward scaling post-training compute.
This process is designed to instill “harness awareness,” where the model is specifically trained to manage its own memory and context within autonomous agent scaffolds like Claude Code or OpenCode.
While the base V2.5 model is trained to reason across modalities, the Pro version is trained to sustain coherence across more than a thousand sequential tool calls.
The standard V2.5 model balances local and global attention to maintain multimodal perception. The Pro model, however, utilizes an increased hybrid attention ratio—evolving from the 5:1 ratio of previous generations to a more aggressive 7:1 ratio.
This allows the Pro model to “skim” the vast majority of its context while applying high-density attention to the specific 15% of data most relevant to its current objective, a critical feature for debugging large repositories or optimizing graduate-level circuits.
Finally, while both models undergo Reinforcement Learning (RL) and Multimodal Preference Optimization (MOPD), the objectives of these stages differ.
For MiMo-V2.5, the RL stage is used to sharpen perception and multimodal reasoning. For MiMo-V2.5-Pro, RL is focused on instruction following within agentic scenarios, ensuring the model adheres to subtle requirements embedded deep within ultra-long contexts and recovers gracefully from errors during autonomous execution.
This results in the Pro model’s “self-correcting” discipline, as seen in its ability to diagnose and fix regressions during the 4.3-hour SysY compiler build.
In a move that distinguishes it from many “open” models that include restrictive “Acceptable Use” policies, Xiaomi has released MiMo-V2.5 under the MIT License.The MIT License is the gold standard of permissive software licensing. For developers and enterprises, this means:
No Authorization Required: Companies can deploy the model commercially without seeking explicit permission from Xiaomi.
Continued Training: Developers are free to fine-tune the model on proprietary data and even release those derivative weights.
Unrestricted Commercial Use: There are no revenue caps or user-base limits that often plague “community” licenses.
By choosing MIT over a custom “open weights” license, Xiaomi is positioning MiMo as the foundational infrastructure for the next generation of AI agents, effectively inviting the global developer community to treat the model as a public utility.
Xiaomi’s pivot toward frontier AI agents is the logical culmination of a decade spent building one of the world’s most dense hardware-software flywheels.
Founded in 2010 as a smartphone disruptor, the Beijing-based company has executed a high-stakes transition into a vertically integrated powerhouse defined by its “Human x Car x Home” strategy. This ecosystem now encompasses over 823 million connectable smart devices unified under the HyperOS architecture.
The company’s 2024 entry into the automotive sector with the SU7 and the subsequent high-performance YU7 SUV served as a proof of concept for this integration, positioning Xiaomi as a direct competitor to global luxury marques.
By investing 200 billion yuan ($29B USD) into foundational R&D for chips and operating systems, Xiaomi has moved beyond consumer electronics assembly; it has become an architect of the “action space,” using its massive hardware footprint as the primary testing ground for the agentic intelligence found in the MiMo-V2.5 series.
The release has been met with immediate “Day-0” support from the broader AI ecosystem. The MiMo team announced that SGLang and vLLM—two of the most popular high-throughput inference engines—supported the V2.5 series at launch.
This was made possible through hardware partnerships with AWS, AMD, T-HEAD, and Enflame, ensuring the model can run efficiently on everything from cloud-based H100s to domestic Chinese accelerators.
Fuli Luo, the project lead at Xiaomi MiMo and a former key member of the DeepSeek team, underscored the philosophy behind the release on X (formerly Twitter):
“A model’s value isn’t measured by rankings alone — it’s measured by the problems it solves. Let’s build with MiMo now!”
To kickstart this building phase, Luo announced a 100-trillion free token grant for builders and creators. This massive incentive is designed to lower the barrier to entry for developers who want to experiment with the 1M context window without immediate financial risk.
The launch arrives at a critical juncture for AI economics. The shift toward usage-based billing marks the definitive end of the “all-you-can-eat” buffet era for AI services, a trend underscored by GitHub’s announcement today that its AI coding assistant Github Copilot will transition all plans to metered, token-based credits.
As seat-based predictability gives way to consumption-driven costs, premium agentic workflows—which can consume millions of tokens in a single reasoning session—are becoming increasingly difficult for enterprises to budget.
User sentiment has turned predictably cynical, with developers lamenting that they will “get less, but pay the same price” as subscriptions convert into finite allotments. This pricing evolution significantly enhances the strategic appeal of the MiMo series. By releasing under a permissive MIT License, Xiaomi allows organizations to bypass the escalating “SaaS tax” and reclaim financial predictability through private deployment.
Crucially, Xiaomi has eliminated the “context tax” for its API. The 1-million-token context window is now billed at the standard rate—1 token = 1 credit for V2.5 and 2 credits for the Pro version—with no additional multiplier. This stands in stark contrast to the industry-wide move toward session-based caps, positioning MiMo as a refuge for cost-sensitive, high-volume development.
The launch of MiMo-V2.5 is more than just a weight drop; it is a declaration of independence for the open-source community.
By matching Claude Sonnet 4.6 in multimodal agentic work and Gemini 3 Pro in video understanding, Xiaomi has proven that the gap between “closed-door” labs and open research is effectively closed.
With the MIT license as a catalyst and a 100T token grant as fuel, the coming months will likely see a surge in specialized, agentic applications built on the MiMo backbone.
Confirming the project’s ambitious trajectory, the team noted they are already training the next generation, focusing on “deeper reasoning” and “richer real-world grounding”. For now, MiMo-V2.5 stands as a testament to the power of sparse architectures and permissive licensing in the race toward functional AGI.
What must Board Directors measure in the new AI world: track training, manage supplier and model risk, add guardrails, and reassess business models in revenue and cost.