Over the past several weeks, teams across Google and academic partner institutions have tackled open problems in math, science, and complex systems engineering that had remained open for years. We are making these agentic tools available in Antigravity’s /teamwork (alongside /boost and other agents) so that researchers and developers can apply them to their own work.
Release: llm-gemini 0.37 Google launched Gemini 4 Hydrogen today - the lowest cost member of the Gemini 4 family. It's actually not too far off the price of its much bigger siblings, Carbon and Argon. I'm finding Gemini 4's pricing really confusing.
Gemini 4 Hydrogen isn't getting a slow, security-conscious release like Carbon and Argon.
OpenAI's safety team systematically investigated why the chains of thought in frontier models like GPT-6 Astra are increasingly evading oversight. Four training steps drive the erosion, including architecture choices and reinforcement learning. The analysis shows that the loss of transparency is not a law of nature but also reflects tradeoffs around capability and cost.
Google enters the decision model market. At AI Engineer, Cloudflare showed which parts of a build failure a classifier could handle. TypeSafe AI showed off their production customers.
NaiveAI could potentially top Artificial Analysis’s speed rankings. Our hands-on test suggests its coding quality holds up against DeepSeek.
Today, we’re making Gemini 4 Carbon available to developers, enterprises, scientists, and consumers. Carbon is our strongest model with frontier performance in cybersecurity capabilities and long-horizon agentic workflows across complex knowledge work, software engineering, and scientific reasoning.
Introducing Our Smallest, Most Efficient Model: Muse Ember Meta AI just launched their entry in the Claude Haiku/GPT Luna category: Muse Ember, which they describe as: [...] our smallest and most efficient model designed for complex reasoning execution and the work agents repeat all day: tool calls, coding subtasks, routing, extraction, and long document parsing.
Release: llm-anthropic 0.31 Anthropic released Fable 5.2 and Mythos 5.2 today. Fable 5.2 is currently ranked first on Artificial Analysis . Anthropic say it "solved two open problems in the physics of black holes", so clearly their internal benchmarks have become more ambitious!
Anthropic is announcing two new models today: Claude Fable 5.2 and Claude Mythos 5.2, which it says are designed to take on "daunting" technical challenges. The company claims the new models show improvements in coding, reasoning, and vision, and can generate outputs faster than the 5.1 versions of the Fable and Mythos models.
Anthropic says the new models "show improvements in areas like coding, reasoning, and vision," as well as in communicating more clearly with users.
Higher-Order Pruning of Experts in Mixture-of-Experts Language Models Introduction Mixture-of-Experts (MoE) models have enabled state-of-the-art LLMs to be scaled to hundreds of billions of parameters, but they come with a fundamental and costly caveat: every expert MLP must occupy memory at inference time, even though the vast majority are idle for any given token.
Thinking Machines' AI model Inkling 2 Small swears during its internal reasoning, 316 times across 2,655 benchmark runs, according to an analysis by Vals AI. The profanity typically shows up late in long runs that fail anyway. But it doesn't work as a warning signal: 96 percent of failed runs contained no cursing at all.
Amazon SageMaker Inference now supports GPU partitioning with NVIDIA Multi-Instance GPU (MIG), which divides one GPU into hardware-isolated partitions so you can run a different model in each or take a right-sized fraction of a newer accelerator. This post shows how two optional API fields place one model per partition, and what we measured against a whole older GPU.
OpenAI is talking about the AGI era, and Nvidia CEO Jensen Huang says AGI has arrived. At the same time, co-creators of today's models are leaving OpenAI, Google DeepMind, and Meta to start their own labs, raising billions with no product to show. Why? They're betting that the current recipe alone won't be enough.
Inside Inkling 2 Small’s training run and METR’s study of AI research at Anthropic.
A quiet day lets us share some behind the senes.
Stanford researchers tested 50 widely cited claims about LLMs against 20 current models. 44 percent of the findings are already "expired," and another 34 percent have weakened significantly. The average lifespan is just 15.7 months. Claims about what models can't do expire fastest.
Yesterday's hot paper was Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights by Rikiya Takehi, Jayden Teoh, Alex Zhang, and Omar Khattab.
Editing Qwen’s weights to reduce reasoning loops, DeepSeek’s super-agent prototype, and the GDPx benchmark.
dbreunig/nesformer Inspired by qwenputer by Divyansh Agrawal, Drew Breunig baked a working emulator for a Nintendo Entertainment System 6502 CPU into the weights of Qwen3.5-4B. You can now play NES ROMs such as Super Mario Bros (very slowly) running inside Qwen!
The first TPU in space is running a Google DeepMind Gemma 4 model and writing cheesy poetry.
NVIDIA Nemotron 3.5 Super VL is a 120B-parameter mixture-of-experts model that pairs reasoning and tool use with multimodal understanding. This post explores its architecture and use cases, then shows how to deploy it on Amazon EKS, EKS Auto Mode, and Amazon SageMaker HyperPod EKS with NVIDIA Dynamo and vLLM.
A new decoy counting benchmark exposes a gap between seeing and reasoning. Elorian’s Glimpse tries to bridge it by thinking with images.
Modern LLMs achieve impressive performance via pre-training and in-context learning (ICL). However, once deployed, they remain largely static, operating with a fixed knowledge cutoff date that leaves them unable to acquire long-term capabilities or adapt to new information.
Former Deepmind researcher Thore Graepel wants his startup Metis Reasoning to bring AlphaGo's search principle to general reasoning. Instead of locking knowledge inside model weights, an explicit knowledge ledger would make every conclusion checkable against evidence. So far, though, there's no working system and no benchmarks.
OpenAI has released GPT-6.1 Terra, a mid-range model that cuts API prices in half compared to its predecessor. Analysts say it sets a new standard for intelligence per dollar, especially for everyday computing tasks and AI-driven computer control.
OpenAI shares model results on the Kakeya conjecture, Exponential Time Hypothesis, and quantum chromodynamics—and how researchers can verify and build on them.
Introducing Kimi K3.1 Pro Kimi K3.1 Pro landed on the Kimi API today - it's not yet open weights, but Moonshot promise "Full model weights will be released by Nov 4, 2026". They're also offering a "Highspeed" version that's 2-3x faster, at twice the price (except for input token cache hits, which are priced the same).
AI models are solving more mathematical problems. A pattern is emerging in their successes that could help us predict which problems they’ll solve next.
The next competitive advantage in enterprise AI has less to do with which model you choose than with what you build around it. That’s an important shift. For the past several years, much of the AI conversation has centered on model capability: bigger models, better reasoning, larger context windows, higher benchmark scores.
For decades, a fundamental question about the difficulty of computing in high-dimensional spaces has resisted proof. New research resolves this question, removing a major theoretical obstacle to understanding the efficiency of randomized algorithms used across science and computing. The Kannan–Lovász–Simonovits (KLS) conjecture was proposed in 1995.
Models specialized for a domain or a task by fine-tuning general open-source models, such as Qwen and Gemma, inherit a vocabulary built for many domains and general applications. In an essay grader we run in production, that vocabulary split the text the model writes most into many tokens.
Inception says Mercury 3 is the most capable diffusion language model released so far. The company claims it runs four to eight times faster than GPT-6 Luna and Claude Haiku 5.5 and beats both in eight of 12 benchmarks. There are no independent tests yet.
Highlights (TL;DR) NVIDIA Kumo Forecast (formerly known as NV-Tesseract Forecasting) is a multi-variate time-series forecasting model that combines a pretrained Transformer backbone with Domain-Aware Representation and Retrieval (DARR), a context-enhanced forecasting mode that blends direct model predictions with patterns retrieved from historical data. This includes interpretability of the results.
Learn how i14 trained and served Laplace 3B, a linear attention model, on AWS Trainium2 using native PyTorch. Its recurrent state stays fixed at 29.5 MB per sequence regardless of context length, avoiding the growing key-value cache of transformer models and keeping serving memory predictable.
Learn how we fine-tuned GPT-OSS-20B for multilingual reasoning on Amazon SageMaker HyperPod, and how to decide between supervised fine-tuning (SFT) and reinforcement learning with GRPO. SFT took reasoning-language accuracy from 22% to 100%, and GRPO added measurable gains where headroom existed.
If you run models on your Mac or use MPS as your primary acceleration tool for computing, there's plenty to look forward to in PyTorch 2.15! FP8 arrives on MPS, language-model decoding gets faster paths through linear layers and softmax and new Metal kernels tackle some surprisingly expensive parts of audio, video and training workloads.
MiniMax's M3.1 Flash jumps 13 points over its predecessor to 42 on the Artificial Analysis Intelligence Index, at half the cost. It also hits 73.7 percent on the DeepSWE v1.1 coding benchmark for under $0.30 per task, matching Deepseek V4.1 Flash 1017 on both performance and price.
Multi-turn agentic workloads reprocess growing conversation context each turn, driving up latency. See how bi-directional KV cache transfer lets prefill and decode nodes reuse previously computed context across turns, cutting time to first token by up to 1.6x for Qwen3-32B and 3.3x for GLM-5-FP8 on Amazon EC2 P5en instances with EFA.