JetBrains Releases Mellum2: 12B Parameter Mixture-of-Experts Architecture Developer-Focused Model

AI News Flash: JetBrains released Mellum2 on June 1, 2026—a 12-billion parameter open-source model based on Mixture-of-Experts (MoE) architecture, but it only activates 2.5 billion active parameters per inference, making inference over twice as fast as models of equivalent scale, significantly reducing deployment costs, released under Apache 2.0 license. Mellum2 isn’t positioned as a replacement for frontier large models, but rather as a ‘focused model’ in multi-model collaboration systems, handling high-frequency lightweight tasks including prompt classification, tool selection, context compression and summarization for RAG pipelines, sub-agent planning validation, and code completion. The model processes only text and code modalities, deliberately excluding multimodal capabilities to keep the architecture lean—particularly suitable for enterprises deploying in private environments to handle internal code and confidential data. Across multiple benchmarks including code generation, reasoning, science, and math, Mellum2 achieves competitive performance among open-source models of similar scale. The technical report has also been published on arXiv (编号 2605.31268), and model weights are available for download on HuggingFace.

2026-06-01 · 2 min · 384 words · Judy

How Google Built Google I/O 2026 Developer Conference with Gemini

AI News Flash: This is the alt text description for a collage image on the Google I/O 2026 related reporting page. The content only involves visual elements, including the Antigravity Coffee Co. flash event, colorful jellyfish images, Timmy TPU video screenshots, and decorative icons such as the ‘AI’ text appearing three times on the left side and twinkling stars, with no substantive information about Google AI technology, product launches, or event schedules.

2026-06-01 · 2 min · 281 words · Judy

Erin Brockovich Slams Data Center Transparency Issues

AI News Flash: Erin Brockovich, the renowned environmental activist famous for exposing Pacific Gas & Electric’s groundwater contamination case (later adapted into a film of the same name), now turns her attention to data center transparency issues, launching a new advocacy campaign. However, the original summary only mentions her ’new mission’ without specifying the targets, demands, or action details—see the original link for more.

2026-06-01 · 2 min · 266 words · Judy

Making Sense of the Debate Over AI Psychosis

AI News: TechCrunch’s Equity Podcast explores whether tech company CEOs are more prone to ‘AI psychosis’—a term gaining traction in the industry to describe execs who develop unrealistic beliefs about AI capabilities due to over-reliance or trust in the technology.

2026-05-31 · 2 min · 407 words · Judy

Testing Google's 24/7 AI Assistant Gemini Spark: More Practical Than Expected

AI News Flash: Google launches Gemini Spark, an AI assistant positioned as a 24/7 automation helper to handle daily repetitive tasks. Features include email inbox summarization and local event planning suggestions, aimed at reducing user time spent on trivial information processing. Hands-on review shows Gemini Spark is actually quite useful.

2026-05-30 · 2 min · 417 words · Judy

Meta Reportedly Developing AI Wearable Pendant Device

AI news flash: According to tech media reports, Meta is developing an AI wearable device, specifically what appears to be a smart pendant that can be worn around the neck (AI pendant). This move shows Meta continuing to extend its AI capabilities into the hardware device field, continuing its strategic bet on AI hardware after its Ray-Ban smart glasses. However, the currently public information is extremely limited…

2026-05-30 · 2 min · 342 words · Judy

Coders are refusing to work without AI — and that could come back to bite them

AI News Brief: Researchers warn that while AI tools enable developers to produce code at a faster pace, the overall quality of code may not improve—and could even decline. The reason for concern is that when developers become reliant on AI autocomplete and generation, their own programming logic skills and debugging judgment may gradually erode. Once low-quality code accumulates to a certain threshold, it will trigger unpredictable cascading issues in future maintenance, scaling, and security. In short, the productivity gains we’re seeing now might be quietly sowing the seeds of long-term technical debt.

2026-05-30 · 2 min · 391 words · Judy

9 demos of Gemini Omni and Gemini 3.5 in action

AI News Flash: Gemini Omni & Gemini 3.5 hero…

2026-05-30 · 1 min · 153 words · Judy

OpenAI Strengthens Societal Resilience with Rosalind Biodefense

AI News Flash: OpenAI officially launches the ‘Rosalind Biodefense’ program, expanding access to the biology-specific model GPT-Rosalind to specific groups. This open access uses a ’trusted access’ mechanism, with eligibility limited to two groups: vetted developers and US government partners advancing biodefense work.

2026-05-30 · 2 min · 423 words · Judy
Get our weekly AI digest:

AI engineering, trading systems, automation — curated weekly. No spam.