Metaがオープン／クローズド両方のフロンティアを揺さぶる：Muse Spark Openと米国オープンウェイトの復活

Meta Shakes Up the Frontiers, Open and Closed was published on Saturday, October 17, 2026. U.S. open weights are back. This has been a week where the balance of American open weight models took a step change back towards the frontier. Meta took the lead of this charge with New Spark Open, which is now the strongest American-made open model by quite some margin. But the addition of Reflections' first model, Beam, is also relevant. Without these two models, there would be 20-plus open weight models built and released by Chinese companies better than the best Western models. At the time of writing this, Mr. L has announced a strong model, Mr. L L 4, with an index score of 38, but its weights were not yet released. There are rumors of a few more strong open weight models coming from other organizations too. In many ways, the gap between the best American and Chinese models has been reduced by roughly half. We've gone from the best American open models being largely irrelevant to consumers who require the cutting edge of performance to them being behind but relevant. Click through to see an image ranking the top open weight models from the China and U.S. model makers on the Artificial Analysis Intelligence

Index. MewSpark Open comes in at 10th behind the likes of Mimo V2.6 Pro, GLM 5.3, Kimi K3, Step 5 Preview, and others. This paints the picture of the cutthroat competitive nature of the frontier model business. Reflection and Mistral announced their model scores, and Meta dropped the weights of a smarter model before they could get their own weights out. All of these models are competing with Chinese models released months ago, where the industry is collectively asking, where have Kimi and Zhipu AI been since the releases of GLM 5.3 and Kimi K3 on August 18th and July 16th, respectively? That is a very nuanced question, which I'll dig into in another time. Today's piece is focused on Meta's turnaround, what MuseSpark Open means for the ecosystem, and the balance of power across Western frontier labs. Why MuseSpark Open should take off and why it may not. One thing we see again and again at InterConnects through our open model analysis is that Western open-weight models get astonishingly more adoption than Chinese models with the same performance to size footprint,

even though the Chinese models usually come out first. MuseSpark Open is landing into this environment at the same time as when open-ended NLM agents are finally real, e.g. GLM 5.2 capabilities and on. Riding off a rapid increase in performance of Chinese models, the open ecosystem has been living through an acceleration in demand. The inference providers are rapidly scaling, the open post-training economy is forming, and countless enterprises are looking to own their own intelligence. USpark Open's performance falls somewhere above GLM 5.2, near DeepSeq v4.1 Flash, and behind GLM 5.3 or Kimi K3. It should be squarely good enough to slot right into this. For some background from the Muse Spark Open model a 816B total 42B active parameter open weights model with a context window of 1 million tokens Built for coding and gentic work MUSE Spark Open is natively multimodal This open weights version has additional safety post trading to make it more robust against misuse as a standalone model without system guardrails.

We recommend deploying MUSE Spark Open as a part of a broader AI system with additional guardrails suited to the specific use case. End quote. Here's the distribution of various Muse models on Artificial Analysis Intelligence Index. Click through to see an image with Muse Wonderful at 56, Muse Wonder at 52, Muse Spark 1.3 at 48, Muse Spark 1.2 at 40, and Muse Spark Open at 39 on the Artificial Analysis Index. This is included to loop back to the original model card snippet. It was shared that this open-weight model underwent additional safety training to prepare for this open release. Many observers saw this release as potentially a model that was broken during safety training as it came out after MuseSpark 1.3. In reality, this is much more likely a safety tune of MuseSpark 1.2, which would only be a one-point drop in artificial analysis to shore up safety behaviors. When Meta released MuseGlimmer, the 30B agentic model, Alex Walling said that they were preparing to release the MuseSpark 1.2 model as open weights. If anything will hold MuseSpark open back, it'll be the specific license they designed for it.

In summary, this custom license has pretty much all the restrictive license behaviors we've seen and criticized here in the last few years of building with open models. They are as follows. One, users of the models and products must display built with Muse on their products. Meta did this with Llama and Kimmy has tried it, but almost no one complies. Two, there is a revenue gating factor where companies deploying the model in a few forms that make more than $20 million per month need a separate license with Meta. Other prominent open model providers are doing this too, in order to set up revenue share agreements for reselling inference. Three, Musepark Open makes you agree to an acceptable use policy, which in the true open source software corner of the world makes this model extremely not open source. Four, all of the above applies to outputs for the model as well, and derivatives, so there's legal ambiguity if you can safely use this model for synthetic data. The license does permit using Musepark Open for internal research and applications. We're at a weird equilibrium where this license is complex and unproven in court, but many players will likely ignore it.

Many, many people who build on this model for various inference providers and meta will only likely go after egregious violations. For now, licenses haven't been the determining factor of AI adoption, but this could change in the future. This is why we think it's an unnecessary strategic risk for labs trying to compete and regain adoption to have complex, non-standard open licenses for their models. Is Meta a Tier 1 OpenAI Anthropic or Tier 2 Gemini XAI Frontier Lab Just a few days before MuseSpark opened Meta unveiled their closed Frontier models Muse Wonder and Muse Wonder Full On traditional benchmarks these are very strong models only behind Anthropics Opus 5.5 and similar to OpenAI's GPT-6 Astra or Google's Gemini 4 Argon. These major releases are coming just over a year out from Meta's massive AI redirection, characterized by CEO Mark Zuckerberg's personal superintelligence post, which got a lot of criticism at the time. Meta is the only American company with Frontier open and closed models.

Alibaba's QN has followed a similar strategy, but with a far more open-leaning approach. Click through to see the benchmarks of MuseSpark Open and MuseWonder. The vibe checks across all of these models have been solid, but the expectations were for them to not immediately challenge the level of refinement of the most recent Anthropic and OpenAI models in real-world usage. I'm excited to try them more. Time to sign up for another AI subscription. I like this recent run of form from Meta, driving their own products and directions in the popular Muse agent, taking wins like the leading U.S. open way crown when useful. Whether or not it's financially successful, it's a unique perspective that fits their history as a business. It often feels like there's a revolving door of which lab is nipping at the heels of the Anthropic OpenAI leading duo. It's not clear that in the long run, Gemini and XAI are out of it, but those two are much more muddled in their offerings than Meta today. If you listen to Zuck and Alexander Wong and the Cheeky Pint, they clearly had a very specific vision of a high-ownership,

high-compute-per-researcher lab that I would expect to pay off. Many labs die due to layers of bureaucracy and misaligned incentives. Meanwhile, Alex Wong has over 200 direct reports, and they're releasing very consistent model and product improvements. Much of my worldview of building models is that you need the resources, the compute, the data, and the talent, all downstream of large sums of money, but organizing them is very hard. Gemini is sometimes described as a place where it's impossible to get any work done or grow, which is embellishing the issue, as many people love it there, but there are kernels of truth to it. Layers of management easily get in the way of making a great model. While I'm praising meta today, the Lama failure has often been described as this sort of management going too far, with model decisions being made out of politics rather than performance. XAI is in its own lane, where it may be closer to having too much talent turnover are not enough organizations to sustain great models for a long period of time. When we look back at the last few years, all of Meta, Gemini, and XAI have produced models pretty much at the frontier of performance, but what has distinguished OpenAI and Anthropic

is their consistency in maintaining frontier status. Anthropic seems the most poised and organized, with a plan and vision for how these organizations should work, and they've gained a lead from it. OpenAI has been more hot and cold, but manages to keep up at the frontier of deployed models while being the leader in public scientific breakthroughs from Navier Stokes to O1 inference time scaling Here the state of play among the companies that this post focused on plus the Xiaomi Viewpoint 2 Pro representing the leading open weight model overall The rankings would be Anthropik above OpenAI, above Meta, which is similar to Gemini, who are above XAI, which is similar to the best Chinese open weights, which are better than the best US open weights. Open questions for the AI industry. After a long time of predicting the frontier model game getting too expensive for many to keep up, we may be starting to see this play out among Western models. We seem very far from one person wins with RSI, given how close all the models are to each other, but we're narrowing down the number of labs who can compete in the U.S. from a much wider net

with surprises, to the five of Anthropic, OpenAI, Meta, Gemini, and XAI. If Microsoft starts releasing models as they're posting strong scaling science results, they could join this list. Given their financial backing with the hyperscale business and direct access to OpenAI's modeling intellectual property. The best models we've seen from the, quote, tier three players like Mistral, Nemetron, Reflection, RC, etc., have had an increasingly hard time impacting either the open or closed frontiers. These models still can get substantial adoption, enough to start growing a business, but the costs of building true frontier models are now measured in billions per major model, and only huge companies can sustain that. Despite this massive scaling in resources, the technical core of the industry still doesn't have a high hit rate on predicting how successful a model will be in real-world usage. Yes, evaluations are a good proxy, but why did the leading labs like Anthropic go from hit models like Cod Fable 5 to total duds in Cod Opus 5, and then back to hits with Opus 5.5?

There are billions of dollars to be made in predicting this accurately, and the labs are surely trying, but the models fundamentally still are largely unknowable in the long tail. Of all the players left in the race, betting on Meta, Gemini, and Microsoft financially seems much easier from a long-term perspective, due to their balance sheets. This is the AI will take a while, and OpenAI or Anthropic do not have a secret sauce perspective. After all, it's the fastest revenue-growing growth industry ever. Competition will naturally be immense. Currently, Meta is valued at under $2 trillion, and Anthropic is approaching their barnstorming RPO at the same price. Meta seems half a step behind on the same technology, but in a long-term view, these seem to be starting from almost the same point, and Meta has the extremely profitable core business. One of the two valuations needs to correct massively. A fun thought experiment I've been pondering with friends is, what do you expect Anthropik's public valuation to be in November of 2027? How about 2028? The longer Meta and the competitors to OpenAI Anthropik stick around, the more negative valuation pressure will exist around AI IPOs.

And that's not to say anything about pricing and safety risks. Time will tell. Markets can stay irrational for a very long time. Thanks for listening and have a great day.