Skip to content
K = 0.731
All news · Today0 items

#Reasoning

All collected and published items, ordered by time. Picks are a selected subset.
First-party

Direct announcements from model vendors, companies and research institutions on their official sites, blogs and accounts.

10/16Fri
  1. Hugging Face Blog83

    Reflection releases Beam, a 501B open model for coding and agentic work

    Reflection released Beam on Hugging Face, a 501B MoE model with 23B active parameters per token, a 1M-token context window, and an Apache 2.0 license. The text-only model targets coding and agentic work, with BF16, FP8, and NVFP4 checkpoints and day-0 support in transformers, vLLM, and SGLang. Hugging Face provides deployment and post-training recipes alongside its own task evaluations and demos.

    Why it matters: The deployment guidance distinguishes active computation from total weight memory, helping teams assess hardware requirements for running and adapting a sparse model with a long context window.

  2. Hugging Face Blog74

    Granite Switch 4.1 Combines Chat, Decisions, and Speech in One Endpoint

    IBM released Granite-Switch-4.1-8B and -30B, combining chat, a general-purpose Decision aLoRA, a specialized Guardian aLoRA, and Granite-Speech-5.0-Turbo-CTC transcription through one hosted endpoint. IBM says the combined model uses approximately the same GPU VRAM as its base model, with adapters reusing the base model’s KV cache. Users can also compose custom Switch models from Granite Libraries adapters; an Adapter SDK for training adapters in Mellea is planned.

    Why it matters: Shared KV caches and modular adapters offer a concrete approach to combining general and specialized decisions with transcription while keeping GPU memory use close to the base model.

  3. Microsoft Research79

    Microsoft Research presents Magnetar for proof-based general-purpose computing

    Microsoft Research presents Magnetar, a zero-knowledge virtual machine that proves RISC-V program execution using security based only on standard hash functions. It supports programs written in Rust, Go, or C without requiring developers to rewrite them as cryptographic circuits. Across 15 benchmark programs, Microsoft reports average speeds 10 to 39 times faster than the fastest existing comparable zkVMs on the same machine, with proofs verified in about 12 milliseconds.

    Why it matters: The benchmark comparisons connect proving speed with general-purpose execution and standard hash functions, clarifying the tradeoffs that have limited broader adoption of verifiable computing.

10/15Thu
  1. Mistral AI77

    Mistral AI releases Timestral, a family of time-series foundation models

    Mistral AI released Timestral in four sizes, from 5M to 209M parameters, for probabilistic, multivariate forecasting with past-only and known-future covariates. Mistral reports that its 209M model ranks first among pretrained, zero-shot models across five metrics on GIFT-Eval, TIME and fev-bench, using a single checkpoint without ensembling. All sizes support CPU inference, and Mistral published a GitHub repository for running Timestral on users’ data; domain adaptation and fine-tuning are available through Mistral Forge.

    Why it matters: The comparisons across model sizes and CPU latency provide a concrete basis for assessing forecasting accuracy against compute requirements in production deployments.

  2. Hugging Face Blog74

    Apertus 1.5 Release Notes Detail Multimodal Understanding, Thinking Mode, and Longer Context

    Project Apertus released Apertus 1.5 in 8B and 70B sizes, adding image and audio understanding, optional thinking, a 262,144-token context window, and improved instruction-following and tool use. The weights were released on Hugging Face on July 24, 2026 under Apache 2.0; the models generate only text, and spoken-language processing remains experimental. The team also publishes training data, recipes, intermediate checkpoints, and reproducible evaluation records.

    Why it matters: Published intermediate checkpoints, training mixtures, and evaluation settings provide a concrete basis for investigating how individual training stages change model behavior and for reproducing reported results.

  3. Google Research78

    Google releases Deep Review for analyzing complex technical documents

    Google Research has publicly released Deep Review, an agentic AI system for technical verification and feedback on research papers and other complex documents, including legal and financial materials. Evolving from the Paper Assistant Tool, which provided optional feedback on more than 16,000 manuscripts, it combines document segmentation, adaptive compute budgets, parallel verification, and global synthesis. Users worldwide receive two complimentary reviews per day, with higher limits and API integration available through a Gemini API key; uploaded documents are never used to train, fine-tune, or improve Google models.

    Why it matters: The review pipeline offers a transferable approach to checking long documents by combining section-specific compute budgets with full-document context and final consistency checks across sections.

  4. Google Research78

    Google evaluates Earth AI geospatial agents and expands experimental access

    Google released a technical report evaluating its experimental Geospatial Reasoning agent across four benchmarks and is expanding early access through Ask Google Earth and an Agent-to-Agent API. Evaluations using Gemini 3.8 Flash and Gemini 3.1 Pro found that specialized Earth AI tools improved spatial reasoning, while upfront planning helped with open-ended workflows. Google is releasing development splits for GeoKTools and GeoExpertQueries, but cautions that GeoExpertQueries evaluates adherence to expert workflows rather than final-answer correctness, making agent outputs drafts for expert review.

10/14Wed
  1. Hugging Face Blog79

    Liquid AI releases LFM2.5-24B-A2B for fast, private on-device agents

    Liquid AI released LFM2.5-24B-A2B, an open-weight reasoning model with 24B total parameters and 2B active parameters, for on-device agents that reason and call tools. Its context window expands from 32K to 131K tokens, and both the reasoning and Base versions are available on Hugging Face. Liquid AI reports 184 tokens/s on an Apple M5 Max and competitive tool-use results, while acknowledging that larger models retain advantages on knowledge-heavy and coding benchmarks.

    Why it matters: The benchmark comparison helps distinguish workloads suited to local deployment, with strong tool use and instruction following alongside weaker knowledge and coding results than some larger models.

  2. Google DeepMind94

    Google DeepMind launches Gemini 4 Argon for complex reasoning and agentic tasks

    Google DeepMind has launched Gemini 4 Argon globally through its API and selected Google products, with access depending on subscription or service tier. Introductory pricing is $2 per million input tokens and $10 per million output tokens, with cached input tokens at 95% off and a 1M unified input-output token limit through the Gemini API. Google reports improvements across coding, professional workflows, scientific research, and robotics, alongside stronger agent safeguards and prompt injection resistance.

    Why it matters: Google’s reported gains in code submissions and reductions in rollbacks connect Argon’s engineering capabilities to delivery speed and code stability in internal workflows.

  3. Google DeepMind82

    Google DeepMind Introduces Gemini 4 Argon for Robotics

    Google DeepMind introduces Gemini 4 Argon, a general-purpose physical reasoning model for robot control, simulation reconstruction, physical understanding, and self-improvement. Google DeepMind reports an 87% ERQA score using pure reasoning and a simulation-generated training example that raises real-world robot success from 56% to 75% while reducing completion time from 35s to 24s. Argon is available today to Google AI Ultra subscribers through Google AI Studio and Google Antigravity, and through the paid Gemini API tier using gemini-4-argon in the Interactions API.

    Why it matters: The simulation-to-hardware example provides a concrete basis for assessing how generated training episodes can improve robot task success and completion speed within a reported setup.

10/13Tue
  1. AWS Machine Learning Blog75

    Build long-horizon agents with Grok 4.7 on Amazon Bedrock

    AWS demonstrates how to build and evaluate a tool-calling agent using Grok 4.7 on Amazon Bedrock, comparing four reasoning effort levels across three synthetic tasks. The agent answered 67 of 72 runs correctly; xhigh achieved 18 of 18, while low achieved 15 of 18 using about an eighth of the output tokens. The guide covers task design, automated grading, production monitoring, and operational controls, noting that these small-scale results are indicative rather than statistically conclusive.

    Why it matters: Tasks with provably unique answers and repeated runs provide a reusable way to compare reasoning effort, while separate correctness and token measurements make the trade-off concrete.

10/9Fri
  1. NVIDIA Blog59

    Into the Omniverse: How Developers Turn Ideas Into Simulations With Frontier AI Agents

    NVIDIA explains how its developers use frontier AI agents, including GPT-6 Astra, with Omniverse libraries to build and refine simulations through natural-language instructions, human review and iterative testing. Examples span warehouse scaling, cable routing, robot asset preparation, autonomous-driving workflows and sensor-based digital twin refinement. One warehouse simulation sustained 30 frames per second with 80 conveyors, 40 arms and 800 boxes during a two-minute test; a robot CAD example used estimated physical properties that were not calibrated against the real robot.

10/7Wed
  1. Hugging Face Blog62

    Nemotron fine-tuning reaches gold-medal level at both IOI 2026 and IMO 2026

    The NVIDIA team started with Nemotron 3 and used supervised fine-tuning, reinforcement learning, and a generate-verify-improve reasoning process to build specialized models that reached gold-medal thresholds at both IOI 2026 and IMO 2026.

    Why it matters: The source gives a complete recipe for fine-tuning Nemotron 3 to gold-medal-level results at IOI and IMO, which readers can use to reproduce its data and reasoning workflow.

10/3Sat
10/2Fri
  2. AWS Machine Learning Blog32

    How Amazon Payments uses contextual multi-armed bandits on Amazon SageMaker AI to improve acquisition funnel conversion

    Amazon Payments used a multi-objective contextual multi-armed bandit (LinUCB) on Amazon SageMaker AI to personalize content in its product-acquisition funnel. In a seven-week online A/B test, one audience segment achieved a high-single-digit relative increase in final funnel conversion, while another showed no improvement. The team ran one LinUCB model for each stage—application start, submission, and approval—and linearly combined their UCB scores to optimize the full funnel. It open-sourced a repository that can be tried with synthetic data.

10/1Thu
  1. NVIDIA Blog22

    How NVIDIA AI factories maximize ROI through high output, long life, and broad applicability

    NVIDIA says AI factory return on investment depends on three factors: output capacity, useful life, and demand. Its platform uses full-stack co-design to maximize throughput per megawatt. SemiAnalysis AgentX data show Vera Rubin NVL72 delivers more than 30 times the throughput per megawatt of GB300 NVL72 and, on DeepSeek V4 Pro, can reduce cost per million tokens by up to 45 times.

  2. Google DeepMind88

    Google DeepMind releases frontier model Gemini 4 Argon

    Google DeepMind released frontier model Gemini 4 Argon, initially making it available through the Fairwind Program to trusted cybersecurity defenders before gradually expanding access. Pricing starts at $2 per million input tokens and $10 per million output tokens.

    Why it matters: Gemini 4 Argon raises the maximum output to 1M tokens and reports multiple benchmark results and internal deployment data, offering a view of its limits on long-running workflows.

9/26Sat
9/24Thu
9/16Wed
  1. NVIDIA Blog62

    NVIDIA Vera Rubin NVL72 makes its MLPerf Inference v6.1 debut with leading performance

    NVIDIA submitted preview results for Vera Rubin NVL72 for the first time in MLPerf Inference v6.1, reporting up to 3.7× the throughput of GB300 NVL72 on Qwen3-VL and up to 2.5× on DeepSeek-R1.

    Why it matters: MLPerf Inference v6.1 reports throughput, scaling efficiency, and software-optimization gains for Vera Rubin NVL72 versus GB300 NVL72, offering a basis for judging inference infrastructure value.

  2. Google Research36

    Google Research proposes Retrieve-for-Train: compile diffusion models with RL to bypass inference bottlenecks in complex AI search

    Google Research proposes Retrieve-for-Train, a framework that uses offline RL to discover reward-aligned query fan-outs and compile them into supervision, then distills them into a 53.9M-parameter diffusion retriever. At inference, it generates a complete result set in one non-autoregressive pass without chain-of-thought tokens. The method fine-tunes fan-out language models based on Gemma3-4B and Qwen3-4B, using set-level attribute rewards to address paraphrase collapse in zero-shot LLMs and autoregressive latency.

  3. NVIDIA Blog72

    NVIDIA shares efficiency progress for Vera Rubin and DSX at AI Infra Summit

    At the AI Infra Summit, NVIDIA announced several advances in Vera Rubin and its DSX platform, shifting AI infrastructure measurement from peak performance to useful agentic tokens per megawatt.

    Why it matters: The article reports several measured token-output figures per megawatt for Vera Rubin and DSX, showing how efficiency metrics for AI factories are changing.

9/8Tue
9/3Thu
8/25Tue
  1. Hugging Face Blog62

    IBM releases Granite 4.2 reasoning models in 3B, 8B, and 30B sizes

    IBM's Granite team has released Granite 4.2, its first family of dense decoder-only reasoning models, in 3B, 8B, and 30B sizes. They were post-trained from the Granite-4.1 base and are all licensed under Apache 2.0.

    Why it matters: The article discloses Granite 4.2's architecture, SFT data mix, and multi-stage RL curriculum in detail, offering a look at concrete post-training methods for reasoning models.

  2. Hugging Face Blog60

    Multiverse Computing proposes Quantization-Aware Healing: 4-bit compressed model beats its bfloat16 original on seven of nine benchmarks

    Multiverse Computing has published a paper proposing Quantization-Aware Healing (QAH). After compressing GPT-OSS 120B to 60B parameters and quantizing it to MXFP4, the method distills directly from the original model before compression rather than from a recovered bfloat16 checkpoint.

    Why it matters: The paper presents a repair recipe that distills from the original model after compression and quantization, with stability comparisons against QAT.

8/21Fri
8/13Thu
  1. Microsoft Research36

    MindTopo exposes spatial-reasoning gaps in multimodal LLMs

    Microsoft Research has introduced MindTopo, a benchmark for topological reasoning in multimodal LLMs. It covers connectivity, separation, ordering, containment, and knot relations, across static reasoning and interactive planning. Current models are substantially better at static recognition than interactive planning, but both remain far below human performance; failures occur more often in planning than perception. Image and video generation helps only when topology is preserved within a single frame and remains unreliable across multi-step operations.

8/12Wed
  1. Google Research60

    Google Research proposes knowledge profiling: retrieval, not encoding, is the bottleneck for frontier LLM factuality

    Google Research has proposed a knowledge-profiling framework that distinguishes factual errors caused by encoding failures from those caused by retrieval failures, and released WikiProfile, a benchmark containing 2,150 Wikipedia facts.

    Why it matters: By separating encoding failures from retrieval failures, knowledge profiling offers a new way to diagnose the sources of factual errors in frontier models.

7/29Wed
  1. Berkeley AI Research62

    From CUDA to MLX: how K-Search brings kernel-optimization experience to Apple Silicon

    Berkeley AI Research and IBM Research have extended the evolutionary kernel-search framework K-Search to MLX and added a structured CUDA-to-MLX translation layer that automatically adapts existing CUDA-kernel knowledge to Apple Silicon.

    Why it matters: The article details the CUDA-to-MLX translation layer and reports two kernel benchmarks, showing how kernel knowledge can be transferred across platforms.

7/26Sun
  1. Berkeley AI Research36

    Berkeley AI Research proposes ABBEL, using belief-state supervision to make LLMs efficient at long-horizon interaction

    Berkeley AI Research has proposed ABBEL, a framework that separates an LLM's recursive summaries into natural-language “belief states” and scores their content with supervised signals to reduce performance loss from context compression in long interactions. On the collaborative-programming benchmark CollabBench, reconstruction-based belief scoring cut the performance gap with a full-context model by about 50%, reduced training from 100 steps to 50, and kept peak context length well below the full-context setup.

7/16Thu
7/15Wed
  1. Hugging Face Blog88

    Thinking Machines releases Inkling family of open multimodal models

    Thinking Machines Lab has released the open-source Inkling family on Hugging Face. Inkling is an MoE model with about 1T total and 41B active parameters, natively supporting image, audio, and text inputs with a 1M-token context. The release also includes Inkling-Small, with 276B total and 12B active parameters.

    Why it matters: The article gives Inkling's architecture, deployment setup, and benchmark comparisons, helping readers assess the practical requirements of an approximately 1T-parameter open multimodal model.

7/10Fri
7/2Thu
  1. Mistral AI71

    Mistral releases Leanstral 1.5 formal-verification model

    Mistral has released Leanstral 1.5, a formal-verification model licensed under Apache 2.0 with 119B total and 6B active parameters. It scores 100% on miniF2F, solves 587 of 672 PutnamBench problems, and sets state-of-the-art scores of 87% on FATE-H and 34% on FATE-X.

    Why it matters: The source reports Leanstral 1.5's formal-verification benchmark results and open-source availability, helping readers assess its usefulness in proof engineering.

6/27Sat
  1. Google Research60

    Google speeds up Gemini Nano on Pixel with frozen multi-token prediction

    Google Research has introduced an architecture that attaches a multi-token-prediction (MTP) head to the frozen Gemini Nano v3 model, speeding up on-device generation without changing the backbone weights.

    Why it matters: Google attaches an MTP head to frozen Gemini Nano v3 and reports concrete methods and gains under on-device memory and energy constraints.

6/25Thu
  1. Google Research37

    Google Research studies how reasoning unlocks LLMs' parametric knowledge

    A paper from Google Research at COLM 2026 finds that reasoning traces can unlock parametric knowledge in LLMs that they could not otherwise access, even for simple single-hop factual questions. Tests on Gemini-2.5 Flash and Pro and Qwen3-32B identify two mechanisms: extra tokens act as a “computation buffer,” and “fact priming” builds a retrieval bridge by generating related facts.

6/11Thu
  1. Google DeepMind71

    Google DeepMind releases DiffusionGemma, generating text up to 4× faster

    Google DeepMind has released DiffusionGemma, an experimental open-source model that uses text diffusion to generate entire chunks of text in parallel. It generates text up to 4× faster on specialized GPUs, exceeding 1,000 tokens per second on a single NVIDIA H100 and 700 tokens per second on an RTX 5090.

    Why it matters: Google DeepMind's open text-diffusion model comes with concrete speed and hardware figures, helping readers assess the feasibility of local interactive inference.