Tencent Hunyuan, with Fudan and Tsinghua researchers, released ExplorationBench, a benchmark that tests whether AI systems can discover rules through experiments in verifiable alien worlds. Across 10 frontier systems, feedback from experiments raised the best AlienCode score to 89.0% after four rounds, while closed-book runs without feedback stayed at 0.5–11.0%.
AI systems are now solving math problems faster than humans can verify the answers, according to the source's headline. The source text is limited to a brief note that OpenAI's math breakthroughs may trigger an identity crisis for mathematicians.
OpenAI published 719 AI-generated math proofs covering 372 result families, after withdrawing 3 for a symbol error. Reports say the release falls short of the AGMAI advisory group's standards, since it uses proprietary models, includes reasoning chains for only 10 manuscripts, and leaves about 42% unformalized. Terence Tao argues that rapidly solving famous problems harms the mathematical community's understanding and collaboration.
Of 1,933 IROS 2026 papers, Robot Learning/Embodied AI appears in about 809, while Navigation/Planning covers 564 and Perception/Vision 556. The article argues large models are being embedded into traditional planning, geometry, and control rather than replacing them. Vision-language-action models are shifting toward efficiency, 3D understanding, memory, and system integration.
According to the excerpt, GPT-6 Astra produced an answer to a fusion problem open for 59 years in 20 minutes and 34 seconds. The physicist who posed the question added only a brief encouraging prompt, and the excerpt does not include the full text.
OpenAI has published 722 math papers generated by an unreleased AI model, posted to GitHub, spanning about 20 mathematical subfields. The model did not fully prove the Riemann hypothesis but proved the quasi-Riemann hypothesis, and it also produced theoretical computer science and partial differential equation results. Many papers include Lean files for computer verification, and OpenAI plans to release more of them.
The original article reports on how AI is changing the mathematics profession, as one mathematician reflects on the shift. The provided text is limited to that single sentence, so no specific models, benchmarks, or figures can be verified.
A month after OpenAI's Millennium Prize solution, the company released findings on more than 300 math problems. The Wall Street Journal says the release was an attempt to win back the mathematics community. Only the excerpt was available, so the source's details on the problems and results are limited.
Xiaomi MiMo-V2.6-Pro is applied to new materials R&D across literature review, molecular design, and automated dry-lab experiments for PFAS capture. The source offers no benchmark scores, speed, pricing, or availability details.
We're testing a new "thinking mode" for our image generation on our Alpha website (alpha dot midjourney dot com). We're finding it boosts prompt accuracy, typography and coherence. We'd love your help testing it on your images and telling us what you think. Thanks! <3
Midjourney is testing a "Thinking Mode" on its Alpha website, where users can click "Rerun (Thinking)" in a job's lightbox to regenerate an image. The company says early tests show gains in prompt accuracy, typography, and coherence, and it is asking users to share feedback in its #ideas-and-features channel. It may later offer the mode broadly or as an option to add more thinking after a job.
Generating more output tokens doesn’t necessarily translate to a higher score. GPT-6 Astra (max) scores 8.6% on ~81k output tokens per task, under half the ~180k of Grok 4.7 (xhigh). Three Claude models generated the most output tokens (~202k to ~562k per task) and score 2.8% to 6.4%.
Once hallucinations are accounted for, Muse Spark 1.3 (max) drops from 26.7% to 8.9%, leaving Grok 4.7 (xhigh) first on the headline metric at 9.4%. Kimi K3 (max) falls from 16.7% to 5.3%, and Claude Sonnet 5.5 (max with fallback) falls from 11.7% to 2.8%. GPT-6.1 Sol (max) declines least, from 7.5% to 6.9%.
Artificial Analysis compared six hallucination checkers on the same deliverables from 20 tasks across eight models. GPT-6 Sol and GPT-6 Luna generally flagged the most material hallucinations, while Claude Sonnet 5.5 and Gemini 3.8 Flash flagged far fewer, with Claude Opus 5.5 falling between Grok 4.7 and Sonnet. The counts reflect checker behavior rather than establishing accuracy or ruling out self-preference.
Interesting to see, given the controversy over the OpenAI release of a series of proofs and what it means for the discipline of mathematics, that at least some of the OpenAI proofs seem to have kicked off extremely rapid iterative advances from a wide community of collaborators.
Lewis Tunstall quotes Kyle Cranmer's post about a paper by Nate Gunnarsson on a non-perturbative approach to chiral fermions in the Standard Model, extending Lüscher's abelian result. The paper's acknowledgments state that OpenAI's GPT-6 Astra model was essential, proposing refinement strategies, writing rewrites of the proof, and carrying out Lean verification.
推荐阅读！ ExploreNet: Learning Where to Explore in Diffusion GRPO ExploreNet is a method introduced by @StellaLisy for learnable exploration in diffusion RL. It learns an adaptive exploration distribution conditioned on the current state, rewarded by the diversity of rollouts, aiming for faster, more targeted learning in diffusion GRPO.
Stanford HAI associate directors Risa Wechsler and Russ Altman, speaking at a Stanford orientation, argued that AI agents can deepen scientific research but must be paired with interdisciplinary collaboration. They stressed rigorous, reproducible methods and clearly measured uncertainty, since convincing AI answers are not enough. They also said labs must weigh agent costs and preserve mentorship so that automation supports human participation in research.
Missed yesterday's demo on adaptive agent environments? Stop by the #COLM2026 Google booth #107 today at 11:00 AM PT to catch Zifeng Wang presenting EnvHarness — a flexible framework enabling co-evolution between LLM agents and their training environments.
Wow, the rate of progress on this result has been remarkable to observe!
Full Blog: https://zyphra.com/our-work/expert-coupling-in-moe-pretraining Full Paper: https://arxiv.org/abs/2610.09372
Wow, this is truly inspiring!!
TypeSafe AI released Jev, a "decision model" that returns a floating-point probability rather than text, which can answer yes/no and multiple-choice questions and score content against criteria. The source cites two browser-extension examples: semantic Ctrl+F search and webpage quality scoring. Simon Willison's criticism is that Jev offers no explanation for its numbers.
Ethereum researcher Justin Drake warned on X that AI-assisted math could, in the worst case, break the signature system used by crypto wallets within months, and urged a "bunker mode" in which users move funds to addresses that have never signed a transaction. Vitalik Buterin agreed but cautioned against moving too fast, saying he has lost more money to botched migrations than to hacks. No one has yet broken the current ECDSA signature scheme in practice.
vLLM v0.31.0 is released with 717 commits from 307 contributors, including 96 first-time contributors. Highlights include DeepSeek-V4.1-Flash support, a vllm preload command that keeps weights in GPU memory across restarts, and Model Runner V2 with draft-model speculative decoding. The release also adds large-scale serving, scheduling, and HiSparse fixes, with full notes linked on GitHub.
UniPat AI's PaperBenchX benchmark found the strongest model, GPT-6 Astra, fully reproduced only 13.98% of 93 real research-paper tasks across 12 scientific fields. Reproduction was judged by regenerating outputs in an isolated environment, with 3,168 expert-verified scoring items. UniPat has open-sourced 12 test tasks and kept 81 tasks closed to preserve long-term evaluation validity.
NVIDIA, Princeton University, and the University of Maryland introduced PivotOPD, an on-policy distillation method that teaches multi-turn LLM agents to recover from their most damaging early mistake. Tested on Qwen3-1.7B and Qwen3-8B students, it posts the best average against 13 baselines on ALFWorld, WebShop, and Search-based QA. It recovers from 72.7% of replayed pivotal mistakes, versus 20.3% for standard OPD, with no added inference cost.
You can either hold crypto and keep being stressed out if a math breakthrough would drain your wallet; or someone stealing your keys would drain your wallet; or someone kidnapping you and forcing you to hand over your keys would drain your wallet Or you can just not hold crypto
OpenAI published 722 math manuscripts covering 372 result groups in a new GitHub repository, openai/math, all produced by an unreleased internal model. The author describes the results as including a near-Riemann hypothesis claim pushed to 0.875, and notes that 25 Fields Medal winners criticized the company's approach to AI math research.
AIWhy it matters: The piece traces how AI math results moved from benchmarks to open problems, offering context on verification and the mathematicians' pushback.
Scott Aaronson reports, based on his sources, that some AI companies have begun discreetly investigating whether their latest internal models can break important cryptographic protocols and primitives. He notes that cryptography is conspicuously absent from OpenAI's list of 376 papers, and the quoted post adds that the US government has censored academic quantum cryptanalysis results.
Updated again.
François Chollet compares AI progress to human learning, noting that 1980s research found programming training improves coding but does not transfer to general reasoning. He argues general intelligence is a fundamental brain property rather than a trainable skill, since domain practice improves only that domain. The post is framed as background for his question whether AI's jagged frontier, driven by math and code via RLVR, reflects general capability or continued human-data bottlenecks.
Epoch gave six models 11 real work tasks from its own operations, including graphic design, data insights, and research design, and graded outputs against employee standards. Fable 5.1 and GPT-6 Astra led on average task performance, reliably handling well-defined work such as coding and computational analysis. The report finds that all models still fail on open-ended judgment, including matching Epoch's standards, designing informative experiments, and generating diverse ideas, so the authors conclude AI cannot yet replace workers at Epoch.
AIWhy it matters: The report separates well-defined task reliability from open-ended judgment failures, which benchmark scores on easily verifiable tasks would miss.
Scott Aaronson on The Mathocalypse. 'The UGC proof invents a completely new bizarre code with a noise test. It's some crazy recursive construction. It's not the long code, not the short code - some alien craziness' I hear a lot of this today. Get used to alien craziness.
The updates in these two posts are only five hours apart. In case you wanted to gauge how fast this ship is currently moving.
Ethan Mollick shares early first-hand accounts from mathematicians grappling with hundreds of AI proofs released by OpenAI. He highlights problems solved in ways no human has yet understood, raising questions about what it means to know something. The linked Scott Aaronson post quotes a researcher, Dana, describing the proofs as unclear and hard to read without AI help, with some possibly verified by a Lean certificate.
Cool demo of Perplexity Decider!
Noam Brown says LLMs have crossed a threshold by surpassing top human experts on some research problems, a jump that makes the recent surge in math results feel sudden. He expects breakthroughs in other domains to follow as models keep improving, though capabilities remain jagged and often still weaker than humans.
Math was bound to fall first. The purer the field the easier it is for AI to crack.