According to the project’s GitHub materials, the software organizes information as a kind of digital “memory palace,” structured into wings, halls and rooms instead of relying only on flat search or compressed summaries.The repository says the app is distributed under an MIT license, runs locally after installation and is designed to preserve conversations in full, rather than leaving an A.I. model to decide what should be remembered.

 

View this post on Instagram

 

A post shared by Milla Jovovich (@millajovovich)

The release drew broader attention because of its benchmark claims. In its published documentation, the team said MemPalace scored 96.6% on LongMemEval in raw mode and reached 100% with a reranking setup, a result presented as a major milestone for A.I. memory systems. Those numbers helped the project spread quickly across developer communities already looking for better ways to give chatbots persistent memory.

But the excitement was quickly met by scrutiny. Developers and online commentators questioned how meaningful some of the benchmark claims were, with debate focusing on methodology, testing conditions and whether some of the comparisons overstated the tool’s advantage. What began as a surprising celebrity-linked code release soon turned into a wider argument over how open-source A.I. projects should present performance claims.

The larger significance may be that MemPalace captures two forces shaping the A.I. industry at once: the growing demand for better memory tools and the growing willingness of developers to publicly challenge ambitious claims in real time. In that sense, the project is not just a novelty tied to a famous name, but part of a deeper conversation about credibility, transparency and competition in A.I. software.

Image: MemPalace

Apache Iceberg has effectively won the table-format wars, and Dremio is using that moment to make a sharper case for its own platform: the hard part now is not choosing an open format, but managing it without adding new layers of cost and complexity. Dremio argues that enterprises embraced Iceberg because they wanted interoperability and less lock-in, and that the format has also become increasingly important for AI-era data architectures that need access to structured, semi-structured and unstructured data in one lakehouse. 

Why This Matters for Users

For users, the promise of Iceberg is flexibility. Teams can keep data in object storage, use multiple engines, and avoid getting trapped inside a single vendor’s proprietary format. But Dremio’s post makes the point that openness brings its own operational tax: Iceberg tables fragment over time, metadata grows, snapshots pile up, and performance can degrade unless engineers actively compact files, tune layouts and schedule maintenance jobs. For many data teams, that means time that should go toward new data products, models or business analysis instead gets spent babysitting tables. 

Dremio’s Competitive Angle Is Automation

That is where Dremio tries to distinguish itself from competitors like Snowflake and Databricks. The company says it was built around Iceberg from the ground up, rather than adding support later, and is pitching itself as the platform that automates the parts of Iceberg management that users least want to do manually. According to Dremio, its platform continuously optimizes physical data layout with Iceberg Clustering, automatically adapts query acceleration through Autonomous Reflections, and handles file compaction, snapshot expiration, manifest rewriting and orphan file cleanup without manual scheduling. Dremio explicitly contrasts that with Databricks, where it says customers still manage optimization jobs themselves, and with Snowflake, where it says automation is more limited for Snowflake-managed Iceberg tables. 

The User Benefit Is Less Maintenance, Faster Queries

The value proposition for customers is straightforward: lower operational overhead and better performance without dedicated maintenance work. Dremio says its autonomous optimization reduces the need for full table rewrites by targeting only degraded regions of data layout, while its reflections system materializes only what is needed based on observed query behavior. The company says this can replace more complex silver-and-gold ETL layering with a more virtualized approach and claims query speeds up to 20 times faster than competing lakehouses on TPC-DS benchmarks. That kind of message is aimed directly at teams that like Iceberg’s openness but miss the more hands-off performance tuning of classic cloud warehouses. 

Interoperability Is Still the Main Strategic Message

Dremio is also leaning hard on openness as a competitive weapon. The company says it co-founded Apache Polaris, an open catalog standard, and argues that this helps customers avoid a new kind of lock-in at the catalog layer. In the post, Dremio says every table it manages is accessible through compatible engines such as Spark, Trino, Flink, DuckDB and Dremio itself. It contrasts that with Databricks’ Unity Catalog-centric approach and Snowflake’s managed-table model. For customers building AI and analytics systems across multiple engines and frameworks, Dremio argues that open access to data and metadata is no longer optional. 

Why Iceberg V3 Could Matter More Than It Sounds

The company also uses the post to highlight Apache Iceberg V3, which it describes as the biggest upgrade since row-level deletes in V2. Dremio says it has already shipped V3 table read and write support, including binary deletion vectors that can make updates and deletes faster and less compute-intensive than older position-delete approaches. It also points to new row-level lineage fields, the VARIANT type for semi-structured data, and nanosecond-precision timestamps as features that make Iceberg more suitable for real-time analytics, CDC pipelines, financial services and IoT workloads. Dremio’s argument is that these are not incremental additions but features that make Iceberg more practical for the next generation of AI-heavy data systems. 

What Dremio Is Really Selling

Underneath the format-war framing, Dremio is really making a broader pitch about the future of the lakehouse. It is saying that openness alone is not enough; the winning platform will be the one that keeps Iceberg interoperable while removing the management burden that often comes with it. That gives Dremio a different position from vendors that support Iceberg but still steer customers toward proprietary catalogs, managed layers or heavier operational involvement. 

Image: Dremio

The investment underscores strong demand for real-time data solutions, particularly from companies pushing into AI and large-scale analytics, positioning ClickHouse alongside competitors like Databricks and Snowflake. The firm’s recent acquisition of open-source platform Langfuse — used for monitoring large language models (LLMs) — reflects the growing importance of observability in AI workflows. ClickHouse’s technology is now widely adopted by major enterprise clients, and the valuation marks one of the most significant funding achievements in the database sector in recent years.

Material by Iva Abadjievа

Image: ClickHouse

In response to growing critique of Microsoft’s deep integration of AI features — including Copilot, Recall, and more — developer “Zoicware” published a script on GitHub aiming to strip out as many of these components as possible. According to reporting by Cybernews, the script targets unpopular or intrusive AI features that some users find distracting or privacy-invasive, though security experts urge caution because such modifications could impact system stability or future updates. 

Material by Iva Abadjievа

Photo by Sunrise King on Unsplash 

In a significant stride towards bolstering software supply chain security, Docker has introduced the Docker Maintained Community Packages (MCP) Catalog and Toolkit. This initiative is designed to offer developers and organizations a repository of secure, high-quality open-source packages maintained directly by Docker, addressing the growing concerns over the integrity and trustworthiness of software components in containerized applications.

The Docker MCP Catalog serves as a curated collection of frequently used open-source packages, selected based on community usage patterns and relevance to contemporary development workflows. By standardizing these images under Docker’s stewardship, the company provides a more trustworthy alternative to packages that are often sourced from less vetted or anonymous contributors.

Complementing the catalog is the Docker MCP Toolkit, an open-source suite of utilities aimed at simplifying the processes of building, testing, and verifying Docker images. This toolkit not only assists developers in reproducing Docker-maintained packages but also empowers contributors to create similarly secure and reliable images tailored to their specific use cases.

“Security remains a top priority for the open source and container ecosystems,”

stated Docker in its official blog post.

“By offering both the catalog and toolkit, we aim to reduce uncertainty in the software supply chain and help teams ship with confidence.”​

Docker’s MCP initiative aligns with broader industry trends emphasizing software supply chain integrity, especially in the wake of high-profile vulnerabilities and dependency attacks. It also supports best practices such as Software Bill of Materials (SBOMs) and digital signing, both of which are integrated into the MCP offerings.

The project is anticipated to see widespread adoption among developers seeking vetted base images, DevOps teams aiming to enhance CI/CD hygiene, and organizations striving to meet compliance standards related to open-source usage.​

Docker encourages community feedback and contributions to the MCP initiative via GitHub, promoting transparency and collaboration in building a more secure container ecosystem.

As threats to the software supply chain continue to evolve, Docker’s MCP Catalog and Toolkit represent a timely and proactive measure towards fortifying one of the most critical layers of modern application infrastructure.

The funding news was shared in an unconventional post on WunderGraph’s official blog, reflecting the team’s transparent and developer-first culture. In parallel, the company issued an official statement to TechCrunch offering additional insight into the strategic partnership with eBay and the broader vision for the product.

“Our investment in WunderGraph’s highly performant open source platform will help boost eBay’s API ecosystem and enable our teams to work faster and smarter in building products that help our sellers thrive,” said Bryan Woodruff, VP of Seller Experience Engineering at eBay.

Founded in 2020 by CTO Dustin Deus, CEO Jens Neuse, COO Björn Schwenzer, and CCO Stefan Avram, WunderGraph has maintained its U.S. incorporation from the outset. While most of the founding team is based in Germany, Miami-based Avram joined in 2022 to provide local leadership in the U.S.

Tackling API Sprawl With GraphQL

GraphQL, initially developed by Meta (then Facebook) in 2012 and open sourced in 2015, is a data query language for APIs designed to make data fetching more efficient. Rather than over-fetching or under-fetching data as is common with traditional REST APIs, GraphQL allows clients to request precisely the data they need. This efficiency has made it a cornerstone of modern software architecture, especially as companies adopt microservices.

However, with the proliferation of APIs, managing and orchestrating them at scale becomes increasingly complex. That’s where WunderGraph steps in. The company originally offered a software development kit (SDK) to help unify disparate APIs—including REST, SOAP, and databases like MySQL. In 2023, it secured a $3 million seed round to build what it dubbed a “GitHub for APIs,” a collaborative platform for discovering and sharing APIs.

At the same time, Apollo was making significant headway in GraphQL Federation, an approach to help multiple development teams collaborate on large, distributed applications. But Apollo’s shift in late 2021 from an open source MIT license to a proprietary Elastic License created an opening for competition.

“Our data showed that some people were really looking for an open source alternative to Apollo Federation,” Neuse told TechCrunch. “We figured our current approach is not working, so let’s just put out an open source alternative to Apollo Federation.”

In response, WunderGraph launched Cosmo in late 2023—a fully open source GraphQL federation platform.

Strategic Alignment With eBay

WunderGraph serves as the main maintainer of Cosmo and offers services like hosting, premium support, and integration assistance for analytics, authentication, and observability. While large enterprises can build their own in-house solutions, many prefer to rely on externally supported products like Cosmo, backed by service-level agreements (SLAs).

This is where the collaboration with eBay becomes pivotal. eBay gains the customization and openness of a flexible GraphQL federation solution, while WunderGraph benefits from eBay’s real-world design input and use case validation.

“I would say we are experts in federation, but we don’t have experience in eBay-scale problems,” Neuse explained. “By having this very close relationship, they taught us everything in terms of how we need to build our product so that it can be integrated into companies like eBay.”

This includes making Cosmo modular enough for companies to adopt only the components they need—something that aligns with enterprise concerns about vendor lock-in. According to Neuse, open source is the only viable path forward for widespread adoption:

“This market needs to be as wide as possible. How can we attract everybody? It must be open source. We cannot limit how people use it.”

Looking Ahead

With $7.5 million in fresh capital, WunderGraph aims to expand its 20-person team and enhance its open source GraphQL federation with tools tailored for collaboration and governance—critical for supporting distributed teams at enterprise scale.

“Open source is the future of API management, and enterprises are demanding transparency, flexibility, and control,” said co-founder Stefan Avram. “We’re building the essential plumbing for the world’s biggest platforms, and this funding allows us to scale while keeping our commitment to open source development.”

NVIDIA has unveiled a suite of cutting-edge technologies to accelerate humanoid robot development, featuring NVIDIA Isaac GR00T N1—the world’s first open and fully customizable foundation model for advanced humanoid reasoning and skills.

The other technologies include simulation frameworks and blueprints such as the NVIDIA Isaac GR00T Blueprint for generating synthetic data, as well as Newton, an open-source physics engine — under development with Google DeepMind and Disney Research — purpose-built for developing robots.

Available now, GR00T N1 is the first of a family of fully customizable models that NVIDIA will pretrain and release to worldwide robotics developers — accelerating the transformation of industries challenged by global labor shortages estimated at more than 50 million people.

“The age of generalist robotics is here,”

said Jensen Huang, founder and CEO of NVIDIA.

“With NVIDIA Isaac GR00T N1 and new data-generation and robot-learning frameworks, robotics developers everywhere will open the next frontier in the age of AI.”

GR00T N1 Advances Humanoid Developer Community

The GR00T N1 foundation model features a dual-system architecture, inspired by principles of human cognition.

• “System 1” is a fast-thinking action model, mirroring human reflexes or intuition.
• “System 2” is a slow-thinking model for deliberate, methodical decision-making.

Powered by a vision language model, System 2 reasons about its environment and the instructions it has received to plan actions. System 1 then translates these plans into precise, continuous robot movements. System 1 is trained on human demonstration data and a massive amount of synthetic data generated by the NVIDIA Omniverse™ platform.

GR00T N1 can easily generalize across common tasks — such as grasping, moving objects with one or both arms, and transferring items from one arm to another — or perform multistep tasks that require long context and combinations of general skills. These capabilities can be applied across use cases such as material handling, packaging and inspection.

Developers and researchers can post-train GR00T N1 with real or synthetic data for their specific humanoid robot or task.

In his GTC keynote, Huang demonstrated 1X’s humanoid robot autonomously performing domestic tidying tasks using a post-trained policy built on GR00T N1. The robot’s autonomous capabilities are the result of an AI training collaboration between 1X and NVIDIA.

“The future of humanoids is about adaptability and learning,”

said Bernt Børnich, CEO of 1X Technologies.

“While we develop our own models, NVIDIA’s GR00T N1 provides a significant boost to robot reasoning and skills. With minimal post-training data, we fully deployed on NEO Gamma — advancing our mission of creating robots that are not just tools, but companions capable of assisting humans in meaningful, immeasurable ways.”

Among the additional leading humanoid developers worldwide with early access to GR00T N1 are Agility Robotics, Boston Dynamics, Mentee Robotics and NEURA Robotics.

NVIDIA, Google DeepMind and Disney Research Focus on Physics

NVIDIA announced a collaboration with Google DeepMind and Disney Research to develop Newton, an open-source physics engine that lets robots learn how to handle complex tasks with greater precision.

Built on the NVIDIA Warp framework, Newton will be optimized for robot learning and compatible with simulation frameworks such as Google DeepMind’s MuJoCo and NVIDIA Isaac™ Lab. Additionally, the three companies plan to enable Newton to use Disney’s physics engine.

Google DeepMind and NVIDIA are collaborating to develop MuJoCo-Warp, which is expected to accelerate robotics machine learning workloads by more than 70x and will be available to developers through Google DeepMind’s MJX open-source library, as well as through Newton.

Disney Research will be one of the first to use Newton to advance its robotic character platform that powers next-generation entertainment robots, such as the expressive Star Wars-inspired BDX droids that joined Huang on stage during his GTC keynote.

“The BDX droids are just the beginning. We’re committed to bringing more characters to life in ways the world hasn’t seen before, and this collaboration with Disney Research, NVIDIA and Google DeepMind is a key part of that vision,”

said Kyle Laughlin, senior vice president at Walt Disney Imagineering Research & Development.

“This collaboration will allow us to create a new generation of robotic characters that are more expressive and engaging than ever before — and connect with our guests in ways that only Disney can.”

NVIDIA and Disney Research, along with Intrinsic, announced an additional collaboration to build OpenUSD pipelines and best practices for robotics data workflows.

More Data to Advance Robotics Post-Training

Large, diverse, high-quality datasets are critical for robot development but costly to capture. For humanoids, real-world human demonstration data is limited by a person’s 24-hour day.

Announced today, the NVIDIA Isaac GR00T Blueprint for synthetic manipulation motion generation helps address this challenge. Built on Omniverse and NVIDIA Cosmos Transfer world foundation models, the blueprint lets developers generate exponentially large amounts of synthetic motion data for manipulation tasks from a small number of human demonstrations.

Using the first components available for the blueprint, NVIDIA generated 780,000 synthetic trajectories — the equivalent of 6,500 hours, or nine continuous months, of human demonstration data — in just 11 hours. Then, combining the synthetic data with real data, NVIDIA improved GR00T N1’s performance by 40%, compared with using only real data.

To further equip the developer community with valuable training data, NVIDIA is releasing the GR00T N1 dataset as part of a larger open-source physical AI dataset — also announced at GTC and now available on Hugging Face.

NVIDIA GR00T N1 training data and task evaluation scenarios are now available for download from Hugging Face and GitHub. The NVIDIA Isaac GR00T Blueprint for synthetic manipulation motion generation is also now available as an interactive demo on build.nvidia.com or to download from GitHub.

The NVIDIA DGX Spark personal AI supercomputer, also announced at GTC, provides developers a turnkey system to expand GR00T N1’s capabilities for new robots, tasks and environments without extensive custom programming.

The Newton physics engine is expected to be available later this year.

All these facts are well known to experienced people in the tech industry and were cited by Mark Zuckerberg in his article ‘Open Source AI Is the Path Forward’ posted on Meta’s blog.

Open-source AI is rapidly closing the gap

According to Zuckerburg, AI is expected to develop in a similar manner. Currently, several tech companies are developing leading closed models. However, open-source AI is rapidly closing the gap. Last year, Llama 2 was comparable to an older generation of models. This year, Llama 3 is competitive with the most advanced models and leading in some areas. Starting next year, future Llama models are expected to become the most advanced in the industry. Even now, Llama leads in openness, modifiability, and cost efficiency.

Llama 3.1 405B, the first frontier-level open-source AI model

Significant steps are being taken toward making open-source AI the industry standard. Llama 3.1 405B, the first frontier-level open-source AI model, is being released along with new and improved Llama 3.1 70B and 8B models. These models offer significantly better cost/performance compared to closed models. The 405B model’s openness makes it ideal for fine-tuning and distilling smaller models.

In addition to releasing these models, there is collaboration with companies to grow the broader ecosystem. Amazon, Databricks, and NVIDIA are launching full suites of services to support developers in fine-tuning and distilling their models. Innovators like Groq have built low-latency, low-cost inference serving for the new models. The models will be available on major clouds, including AWS, Azure, Google, Oracle, and more. Companies like Scale.AI, Dell, and Deloitte are ready to help enterprises adopt Llama and train custom models with their data. As the community grows and more companies develop new services, Llama can collectively become the industry standard, bringing AI benefits to everyone.

“Meta is committed to open-source AI”

Mark Zuckerberg is claiming that Meta is committed to open-source AI. He even outlines why open source is the best development stack, why open sourcing Llama is beneficial for Meta, and why open-source AI is good for the world, ensuring long-term viability.

Why Open Source AI Is Good for Developers

Open source AI is beneficial for developers as it allows for customization of models to meet specific organizational needs without external oversight, provides independence from closed vendors ensuring control over models and flexibility, enhances data security by allowing models to run locally with transparent development processes, offers cost efficiency with models like Llama 3.1 405B running at approximately half the cost of closed models, and represents a promising future investment as open source technology is advancing rapidly and is expected to remain advantageous in the long term.

You can access the models now at llama.meta.com. 

Photo: about.meta.com

“This code holds an important place in history and is a fascinating read of an operating system that was written entirely in 8086 assembly code nearly 45 years ago,” the blog post remembered. The release includes not only the source code for MS-DOS 4.00 but also additional beta binaries, documentation in PDF format, and disk images. These materials have been preserved and made accessible thanks to the efforts of internet archivist Jeff Sponaugle and the guidance of former Microsoft CTO Ray Ozzie.

This version of MS-DOS, developed in collaboration with IBM, has a complex history, featuring contributions to what would eventually evolve into OS/2. Notably, this release includes early, unreleased beta binaries discovered by a young English researcher, Connor “Starfrost” Hyde, who found them among Ray Ozzie’s collection of software.

Microsoft’s Open Source Programs Office (OSPO) explored the possibility of releasing the source code for MT-DOS but eventually focused on MS-DOS 4.00. Although the full source code for MT-DOS was not found, the release of MS-DOS 4.00 represents a rich piece of computing history, showcasing an era when operating systems were written entirely in 8086 assembly code.

The released materials can run on hardware as old as an original IBM PC XT and as recent as a Pentium, and are also compatible with open-source emulators like PCem and 86box, allowing enthusiasts to explore this vintage software in a modern setting.

The initiative underscores the value of digital archaeology in preserving and understanding the technological advancements of the past. Microsoft and IBM’s collaborative effort highlights their ongoing commitment to sharing important historical artifacts with the public and contributing to the educational and technological community.

According to a company blog post, the move is in line with the team’s desire to let users and developers shape the future of artificial intelligence in Windows Terminal by fostering a collaborative environment for innovation.

Terminal Chat, which is currently available in Windows Terminal Canary, allows users to have conversations with an AI service directly in the terminal. This feature enables users to receive intelligent suggestions, such as searching for commands or understanding error messages, while maintaining the context of their terminal session.

The current implementation of the Terminal Chat feature in Windows Terminal requires users to furnish their own Azure OpenAI Service endpoint and key, as it lacks an integrated large-language model. Those keen on utilizing Terminal Chat can locate the corresponding code in the feature/llm branch of the Windows Terminal repository on GitHub. Furthermore, the most recent build of Windows Terminal Canary, inclusive of the Terminal Chat functionality, can be obtained by downloading from the GitHub repository.

Configuring Terminal Chat in Windows Terminal Canary involves the manual addition of an AI service endpoint and key to the Terminal Chat settings. Currently, Terminal Chat exclusively integrates with the Azure OpenAI Service. To acquire the essential Azure OpenAI Service endpoint and key, users must create and deploy an Azure OpenAI Service resource.