Also today
Datalab's Marker 2 Hits 76.0 on olmOCR-Bench at 5x MinerU's Throughput
Datalab has released Marker 2, a document parsing tool that scores 76.0 on the olmOCR-bench and processes documents at five times the throughput of MinerU, outperforming competitors including Docling and LiteParse. This positions Marker 2 as a leading open-source option for high-volume document ingestion pipelines, a common bottleneck in RAG and enterprise AI workflows. The benchmark results matter because olmOCR-bench tests real-world document understanding across complex layouts, making the score a meaningful signal of production-readiness. Developers building document-heavy applications — legal, finance, healthcare, or general knowledge base construction — should benchmark Marker 2 against their current stack. The throughput advantage in particular is significant for teams processing large document corpora where latency and cost are constraints.
MarkTechPost
AlphaFold AI Used to Redesign Gene-Editing Proteins for Improved Safety
Researchers have leveraged DeepMind's AlphaFold to redesign gene-editing proteins, producing variants with improved safety profiles by reducing off-target activity. The work demonstrates that AlphaFold's structural prediction capabilities extend meaningfully into protein engineering — not just prediction — enabling targeted modifications that would be extremely difficult to achieve through traditional experimental methods. This is a significant proof point for AI-assisted biological design, showing that foundation models trained on structural data can guide consequential therapeutic development decisions. For developers and engineers working in biotech or computational biology, this underscores AlphaFold as an active design tool rather than a passive lookup system. It also continues to validate DeepMind's long-term investment in structural biology AI as having real downstream scientific and medical impact.
Google DeepMind
Meta Updates Its AI Chatbot to Focus More on Productivity and Assistant Features
Meta is updating its AI chatbot to more closely resemble a general-purpose productivity assistant, shifting emphasis away from conversational interaction and toward task completion and utility. The update reflects a broader strategic move by Meta to position its AI as a daily-use tool embedded across its platforms, including WhatsApp, Instagram, and Messenger. For developers building on or alongside Meta's AI ecosystem, this signals an increasing focus on agentic and task-oriented interaction patterns at scale. The shift also increases competitive pressure in the assistant space, where Meta's massive user distribution gives it a structural advantage over standalone AI apps. Engineers watching platform AI integrations should track how Meta's assistant APIs and capabilities evolve as this productivity pivot accelerates.
Meta AI
NVIDIA and Partners Outline South Korea's AI Infrastructure Roadmap at AI Summit
At a dedicated AI summit in South Korea, NVIDIA and local partners outlined a coordinated roadmap for AI infrastructure expansion across the country, covering data center buildout, model deployment, and sovereign AI development goals. The event underscores NVIDIA's strategy of embedding itself into national AI plans as a foundational hardware and software partner, extending beyond individual enterprise deals. For developers in the Asia-Pacific region, this signals meaningful investment in local compute availability and AI services infrastructure over the coming years. The partnership framework also reflects a growing trend of countries pursuing sovereign AI capacity rather than depending entirely on US-based cloud providers. Engineers and teams evaluating infrastructure for regional deployment should watch Korean cloud and compute partnerships as they formalize.
NVIDIA
AI Firms Push for More Data Centers as EPA May Reduce Community Oversight
A regulatory shift under the Trump administration's EPA may reduce the ability of local communities to challenge or delay data center construction permits, a change that would directly benefit major AI companies racing to expand compute infrastructure. AI firms have been vocal about infrastructure bottlenecks as a limiting factor on model training and inference scaling, making permitting reform a meaningful unlock for the industry. The policy change is primarily framed around environmental review processes, which have historically been used to slow or block large industrial facilities including data centers. For developers and engineers, this represents a potential acceleration in US compute capacity availability over the next two to three years, with implications for cloud pricing and GPU availability. The move is also likely to face legal and political challenges from environmental groups and local governments.
Ars Technica
Midjourney Acquires Astrology App Co-Star in Unexpected Strategic Move
Midjourney has acquired Co-Star, the popular AI-powered astrology app, in a deal that marks the image generation company's first known major acquisition. The move is surprising given Midjourney's narrow focus on generative image capabilities, and the strategic rationale is not yet fully explained — possibilities include user data, consumer app distribution, or a pivot toward broader AI consumer products. Co-Star has tens of millions of users and is one of the few consumer AI apps with strong daily engagement, making it a potentially valuable distribution asset. For developers watching the generative AI space, this acquisition hints that Midjourney may be building toward a broader consumer AI platform rather than remaining a single-purpose creative tool. The deal is worth tracking as a signal of where second-generation AI companies are investing beyond their core product.
The Verge
Building an End-to-End OCR Pipeline with Baidu's Unlimited-OCR for High-Resolution and Multi-Page PDFs
A new technical guide details how to construct a production-ready OCR pipeline using Baidu's Unlimited-OCR library, specifically targeting high-resolution image inputs and multi-page PDF parsing — two scenarios where many open-source OCR tools degrade significantly. The walkthrough covers pipeline architecture from ingestion through text extraction and post-processing, making it directly actionable for developers building document understanding systems. Unlimited-OCR's ability to handle high-resolution inputs without tiling artifacts or context loss is a meaningful differentiator for document-heavy applications in legal, finance, and enterprise data extraction. This is particularly relevant for teams assembling RAG pipelines or knowledge bases that depend on accurate structured extraction from heterogeneous document types. Developers evaluating OCR components should test Baidu's tooling alongside Marker 2 and other recent entrants given the active competitive landscape.
MarkTechPost
AI-Based Clinical Decision Support System Shows Efficacy for Inherited Retinal Disease Diagnosis in Randomized Trial
A multicenter randomized trial published in Nature Medicine evaluated an AI-based clinical decision support system for diagnosing inherited retinal diseases, finding that the system meaningfully assisted clinicians in making accurate diagnoses in a controlled setting. The trial design — multicenter, randomized — is notably rigorous by clinical AI standards, giving the findings more credibility than retrospective or single-center studies. Inherited retinal diseases are a diagnostically challenging domain due to their genetic heterogeneity and the expertise required to interpret imaging and clinical data together. For developers working in medical AI, this represents a meaningful benchmark for how decision support systems can be validated for clinical deployment. The publication in Nature Medicine also signals growing acceptance of AI clinical tools in top-tier medical literature.
Nature.com