iiinova — where your digital self keeps evolving

hardmaru @hardmaru.bsky.social · 28/09/2026

Our Neuroevolution textbook is finally in print! Free online edition: neuroevolutionbook.com Pre-order: a.co/d/06DpkH9h I am incredibly grateful to my co-authors Sebastian Risi, Yujin Tang, and Risto Miikkulainen for making this happen. Thread:

2369

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 28/09/2026

Introducing "Scaling In-Context Imitation Learning" (SAIL) to be presented at #IROS2026. This work is a collaboration between Sakana AI and the University of Tokyo. Blog: pub.sakana.ai/sail Paper: arxiv.org/abs/2603.08269 What does a robot need before it can tackle a new task? Thread:

2144

hardmaru @hardmaru.bsky.social · 26/09/2026

Follow Sakana AI on IG → SakanaAILabs

071

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 26/09/2026

Sakana AI is on Instagram now. www.instagram.com/sakanaailabs

022

hardmaru @hardmaru.bsky.social · 24/09/2026

Jürgen Schmidhuber pioneered meta-learning, recursive self-improvement, and world models back in the 1990s, when compute was a million times more expensive. He has been thinking about machines that improve themselves since before compute was cheap enough to make it practical.

2121

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 24/09/2026

Sakana AI welcomes Jürgen Schmidhuber as Chief Scientific Advisor. Blog: sakana.ai/schmidhuber/ Sakana AI is incredibly proud to announce that Jürgen Schmidhuber, universally recognized as the father of modern AI, is officially joining Sakana AI as Chief Scientific Advisor.

1364

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 17/09/2026

🐙 Fugu Max is now on Sakana Chat 🐙 Free for all to use. Try it: chat.sakana.ai Blog: sakana.ai/chat-fugumax... 🐡

2256

hardmaru @hardmaru.bsky.social · 20/09/2026

日経新聞による、グーグルディープマインド東京を率いる全炳河（Heiga Zen）氏の素晴らしい特集記事。2018年、Heigaさんと2人でGoogle Brain東京チームを立ち上げた日々を懐かしく思います。当時は時差の厳しい深夜の会議をこなしながら、日本のAI研究の存在感を示すために必死でした。現在、彼がGDM Tokyoを率い、私が Sakana AI を起業して、東京のAIエコシステムがここまで大きく成長したことを本当に嬉しく思います！ www.nikkei.com/article/DGXZ...

nikkei.com

GoogleのAI部門支える全炳河氏　生成AIの到来、デビッド氏はサカナに - 日本経済新聞

気がつけば英国に来てから10年がたっていました。WaveNet（ウェーブネット）で燃え尽き、新しいチャレンジを求めていました。家族のことも考えるとそろそろ潮時ではないか。そう考えて日本で人工知能（AI）の研究開発をさせてくれそうな会社をいくつか受け始めました。その当時のグーグルは日本ではほとんどAI研究開発をしていなかったのです。米テック大手の日本法人やAIスタートアップなど複数社からオファー

081

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 16/09/2026

🐡 Sakana Fugu Max is now on Chat 🐡 Sakana Chat just got another big upgrade, following our August update. Two things shipped today: ・Sakana Fugu Max is now integrated, in addition to Namazu ・Memory feature is here Try Sakana Chat: chat.sakana.ai Read more: sakana.ai/chat-fugumax...

0164

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 14/09/2026

Introducing PC-ALM: a local-learning alternative to backprop that trains 1000-layer neural nets using only local dynamics. Blog: pub.sakana.ai/pc-alm/

18714

hardmaru @hardmaru.bsky.social · 12/09/2026

Do large language models actually understand the world, or are they just very good at pretending? Does it even matter? Our recent Royal Society special issue explores this: royalsocietypublishing.org/rsta/issue/3...

2355

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 12/09/2026

Peak performance across hard benchmarks: • Best on 5/8 benchmarks (DeepSWE, Chartography, Toolathon, GDP.pdf, SWEFish) • Chartography: 48.3 (outperforming Opus 5 & Fable 5) • DeepSWE: 74.3 Achieved without Fable 5, Fable 5.1, or GPT-6-Astra in the agent pool. More Info: sakana.ai/fugu-max-rel...

021

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 12/09/2026

Fugu Ultra v2 is now live on OpenRouter! 🐡 openrouter.ai/sakana/fugu-... Our flagship orchestration engine built for peak performance on complex multi-step reasoning, autonomous research, and full-stack software development.

openrouter.ai

Fugu Ultra v2 - API Pricing & Providers

Fugu Ultra v2 is the higher-performance model in Sakana AI's Fugu family. $5 per million input tokens, $30 per million output tokens. 1,000,000 token context window, maximum output of 128,000 tokens.

1111

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 11/09/2026

Fugu Max is live on Open Router! 🐡 Our learned multi-agent orchestration engine routes tasks across open-weights and specialized models for Pareto-frontier efficiency. • $2/$6 per 1M tokens • Multimodal (Image/PDF) + Web Search • Configurable reasoning & Function Call openrouter.ai/sakana/fugu-...

openrouter.ai

Fugu Max - API Pricing & Providers

Fugu Max is the cost-performance model in Sakana AI's Fugu family. $2 per million input tokens, $6 per million output tokens. 1,000,000 token context window, maximum output of 128,000 tokens.

1211

hardmaru @hardmaru.bsky.social · 11/09/2026

Introducing Fugu Max and Fugu Ultra v2: Orchestrating the Pareto Frontier The AI industry has spent a decade optimizing along a single axis: build bigger, more expensive models. But intelligence has never been a monolith. Blog: sakana.ai/fugu-max-rel...

2505

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 11/09/2026

Introducing Fugu Max and Fugu Ultra v2: the next evolution of Sakana Fugu’s multi-agent orchestration system. Try: sakana.ai/fugu Blog: sakana.ai/fugu-max-rel... The frontier that actually matters is the Pareto frontier: capability on one axis, cost on the other.

1323

hardmaru @hardmaru.bsky.social · 30/08/2026

People ask what Japan needs to be more innovative. More English? STEM? Daycare? Those are fine, but won’t move the needle. If I had a magic wand, I’d give the country a collective sense of hope. You can be dirt poor and still build. Hope creates change. Change creates more hope.

2522

hardmaru @hardmaru.bsky.social · 29/08/2026

Watching a popular coding tool lose frontier model access really shows why model resiliency matters. In the future, products will continue to work perfectly even if several underlying models go offline. They’ll just route around them.

1394

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 22/08/2026

We're Hiring: Member of Technical Staff (LLM Development) 🐟 sakana.ai/careers/memb... The role: ・Developing LLMs that form the foundation of Sakana AI's products, solutions, RSI research ・Joining a team that covers every stage of LLM development, from pre-training to post-training and evaluation

0102

hardmaru @hardmaru.bsky.social · 16/08/2026

unpopular opinion: gemini is a pretty great model most current baseline models are actually perfectly fine for 99% of everyday work. not everyone needs a state-of-the-art coding model to get things done.

8530

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 14/08/2026

The next frontier of Recursive Self-Improvement is Physical AI. Japan sparked the robotics revolution. We’re expanding our RSI Lab to build world models that allow agentic reasoning systems to recursively self-improve their ability to simulate, plan, and act in the real world. Join us in Tokyo:

sakana.ai

Sakana AI

Member of Technical Staff (RSI Lab)

0352

hardmaru @hardmaru.bsky.social · 13/08/2026

We just pushed a big update to Sakana Chat Free to use: chat.sakana.ai A big motivation for this release is getting people in Japan, especially kids, excited about software development. Vibe-coding turns Kanji & Math drills during the summer break into fun games they can actually build themselves.

0151

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 13/08/2026

Sakana Chat just got an upgrade. Powered by Fugu and Namazu, our Japanese LLM. With newly added code execution, you can vibe-code interactive web apps, games, and tools in seconds by describing them (even in Japanese). No login required, free to use: chat.sakana.ai 🐟

2224

hardmaru @hardmaru.bsky.social · 05/08/2026

After rigorous testing, our joint AI project with Daiwa Securities is entering the full-scale production phase. We're bringing our agentic AI systems to Daiwa’s wealth management teams to accelerate complex market analysis in volatile markets. Big milestone for Sakana AI today!

1161

hardmaru @hardmaru.bsky.social · 03/08/2026

Sakana Namazu: An LLM API with Japanese-vibes! 🎏 Built for Japanese enterprises, featuring frontier-level reasoning and built-in agentic tools. Blog post: sakana.ai/namazu-api#E...

0122

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 03/08/2026

🐟 Sakana Namazu API 公開 🐟 本日、Sakana AIは大規模言語モデル「Namazu」をアップデートし、API「Sakana Namazu（サカナ・ナマズ）」として提供を開始しました。 Sakana Namazu API: sakana.ai/namazu 🐟

132

hardmaru @hardmaru.bsky.social · 27/07/2026

Open ecosystems are the foundation of a healthy AI industry. We have always believed that collective intelligence is the future. Proud that @sakanaai.bsky.social is standing alongside global tech leaders to sign the open-weights letter. 🐟

0292

hardmaru @hardmaru.bsky.social · 26/07/2026

Fugu-Ultra now works with Claude Code 🐡

1324

hardmaru @hardmaru.bsky.social · 24/07/2026

Our team just shipped Fugu-Ultra v1.1! 🐡 By dynamically orchestrating the latest frontier models, we pushed performance up by 7.9 points. We are now beating Fable 5 in complex coding and reasoning tasks without even having Fable 5 in our agent pool. Collective intelligence is the future.

511515

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 24/07/2026

Announcing Fugu-Ultra v1.1 🐡 We’ve been thrilled by the reception to the Fugu model family. Thanks to everyone who tried it, shared feedback, and trusted Fugu with real work. Today, we’re releasing Fugu-Ultra v1.1 → sakana.ai/fugu Upgraded to incorporate the latest frontier models.

1573

hardmaru @hardmaru.bsky.social · 24/07/2026

Schmidhuber being Schmidhuber 🔥

Schmidhuber: I support open-source models distilling what commercial companies distilled for free from the entire internet. I published distillation for free in 1991 in Europe - this was copied in the US and in China (https://x.com/SchmidhuberAI/status/1608870559609421831)

Quoted:
Director Michael Kratsios (US government worker)

We have information that Moonshot AI distilled Anthropic’s Fable for the development of its K3 model.

To do this they developed a sophisticated internal platform to conduct large scale distillation against U.S. models, allowing them to quickly switch between multiple methods of access to avoid detection. Moonshot AI has also acquired GB300-equipped servers and has accessed GB300s in Thailand, likely to train its AI models.

The United States strongly supports the free and fair development of AI, including a thriving competitive ecosystem that spans frontier models, specialized systems, open-source frameworks, and open-weight models. Legitimate AI distillation used to create smaller, more efficient models plays a vital role in this open innovation ecosystem. However, large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology and undermining American research is unacceptable.

1574

hardmaru @hardmaru.bsky.social · 22/07/2026

Incredibly proud of the Sakana AI team. We have developed an orchestration model right here out of Japan that achieves state-of-the-art performance on real-world cybersecurity benchmarks! 🎌 sakana.ai/fugu-cyber-r...

0544

hardmaru @hardmaru.bsky.social · 13/07/2026

Language models and coding agents are great, but there is more to life, and more to AI, than just LLM agents.

3535

hardmaru @hardmaru.bsky.social · 13/07/2026

How do physical systems achieve collective intelligence and self-repair without a central brain? A new paper published in Nature Communications from Sakana AI, IT University of Copenhagen, Autodesk, presents a beautiful realization of biologically inspired robotics: Smart Cellular Bricks. Thread 🧵

37016

hardmaru @hardmaru.bsky.social · 10/07/2026

One of my first journeys in neural networks started over a decade ago with implementing CPPN-NEAT! Back then, I built a clone of ‘Picbreeder’ not only to study the mechanics of neural nets, but to explore the human creativity process itself, and generate some cool abstract art.

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 27/06/2026

Fugu-Ultra is now available on Vercel AI Gateway vercel.com/changelog/sa... ✨

0102

Reposted by hardmaru

Takashi Ishida @tksiia.bsky.social · 26/06/2026

Excited to share CoffeeBench!!☕️☕️☕️ We evaluate LLM agents in a 90-day B2B coffee supply-chain economy spanning farmers, roasters, and retailers, where autonomous firms negotiate, manage inventory, set prices, handle invoices, and manage cash flow. arxiv.org/abs/2606.16613 github.com/sakanaai/cof...

2214

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 24/06/2026

Fugu-Ultra is now live on OpenRouter! ⚡ We share a core vision with the OpenRouter team: the future of AI isn’t a single monolithic model, but the collective intelligence of the world’s best models working together. Try it: openrouter.ai/sakana/fugu-... 🐡

1272

hardmaru @hardmaru.bsky.social · 24/06/2026

Excited to partner with OpenRouter ⚡ Products like OpenRouter Fusion and Sakana Fugu have sparked a serious conversation about dependency and resilience in AI. I believe this is just the start of a massive architectural shift to come in AI development.

0374

hardmaru @hardmaru.bsky.social · 22/06/2026

Human intelligence is fundamentally a collective intelligence. We solve complex problems by participating in a vast cultural network that builds upon ideas across generations. I believe the strongest AI systems will become a collective intelligence, too.

77111

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 22/06/2026

Sakana Fugu: One Model to Command Them All Frontier-level performance without single-vendor dependency. Fugu dynamically orchestrates the world's best models to tackle complex, multi-step tasks. Plug collective intelligence directly into your workflows today with a single API. sakana.ai/fugu/

sakana.ai

Sakana Fugu — Multi-agent System as A Model

One model to command them all

0101

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 22/06/2026

How does it work? Sakana Fugu is itself an LLM, trained to call various LLMs in an agent pool, including instances of itself recursively. Fugu dynamically orchestrates the world's best models to tackle complex, multi-step tasks. Here, Fugu is a multi-agent system that behaves like a single model.

192

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 22/06/2026

Fugu stands shoulder-to-shoulder with leading models like Fable and Mythos across the industry's most rigorous engineering, scientific, and reasoning benchmarks. Read the full blog: sakana.ai/fugu-release Beyond Bigger Models: Why are Orchestration Models the Next Frontier (Thread Below)

1212

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 22/06/2026

Introducing Sakana Fugu: A full multi-agent orchestration system accessible via a single model API. Our ‘Fugu Ultra’ model matches the performance of Fable and Mythos, delivering frontier capability without the risk of export controls. Try it: sakana.ai/fugu 🐡

48921

hardmaru @hardmaru.bsky.social · 09/05/2026

Reproducing all of Jürgen Schmidhuber’s papers (1990-2025) using an AI coding assistant. Cool project by Yaroslav! It even reproduced the “World Models” paper by me and Schmidhuber (2018) using a toy environment, with a full VAE + RNN world model implementation. Project: github.com/cybertronai/...

1466

Reposted by hardmaru

Sakana AI @sakanaai.bsky.social · 09/05/2026

How do we make LLMs faster and lighter? Don’t force the GPU to adapt to sparsity. Reshape the sparsity to fit the GPU! Our latest work with NVIDIA introduces new CUDA kernels & data formats for faster inference and training of sparse transformer language models: Blog: pub.sakana.ai/sparser-fast...

0252

hardmaru @hardmaru.bsky.social · 08/05/2026

Excited to share Sakana AI’s new #ICML2026 paper in collaboration with NVIDIA: "Sparser, Faster, Lighter Transformer Language Models" arxiv.org/abs/2603.23198 This work introduces new open-source GPU kernels and data formats for faster inference and training of sparse transformer LLMs: 🧵 Thread 👇

2618

hardmaru @hardmaru.bsky.social · 03/05/2026

If GitHub were built in: Japan 🇯🇵 China 🇨🇳 North Korea 🇰🇵 The EU 🇪🇺

Source: https://twitter.com/almmaasoglu/status/2050603144686473246

Source: https://twitter.com/jonasfroeller/status/2050288312213004578

https://twitter.com/hardmaru/status/2050783184745545850

https://twitter.com/TheEthanDing/status/2050575592886812993

612718