Verified, then logged.
Sun 22 Nov 2026 · 00:13 KST / 15:13 UTC한국어 | ENSearch /

AI

Frontier and open model releases, price and API changes, papers, benchmarks and safety evaluations.

An over-the-shoulder view of a user comparing templates, an uploaded back-view photo, and purchase bundle options across two generic short-video creation services. AI-generated image

VidToC compares video token compression across three tasks

The effects of video token compression varied by task, and the researchers also compared how the model’s underlying architecture mattered.

Last 7 daysNov 17 – Nov 22

Higgsfield compares 8-second clips made from a single prompt

Higgsfield judged Seedance 2.5 the closest match to its prompt after running each generation option once.

Contrastive learning improves AI jailbreak detection

The method classifies new inputs by whether they are closer to a harmful or harmless prototype, without training a separate classifier.
An illustrative AI evaluation lab scene shows a blurred researcher from behind comparing image outputs across matching workstations connected to a server rack. AI-generated image

DOCK solves 15 robot object placement tasks in a simulation

The researchers trained DOCK using estimates of how far objects held by a robot gripper moved, based on the robot’s joint angles.
Three thermal views accompany a three-dimensional reconstruction of an empty industrial room, showing temperature differences between pipes and walls. AI-generated image

E-prop posts 100% survival in a humanoid robot walking trial

A learning rule modelled on biological nervous systems matched the survival rate of a non-biological baseline.
Monitors at an unoccupied computational physics workstation show heat diffusion and a steepening nonlinear wave alongside subtly differing numerical simulations. AI-generated image

Stored abstracts lift ranking score in paper discovery test

KnowMe Paper Radar uses topic relevance and authors’ research histories to recommend papers and shows the evidence behind each part of the score.
An academic editor viewed from behind compares a manuscript with researcher expertise profiles across two monitors to select potential peer reviewers. AI-generated image

MORSEPose learns to estimate pose without surface models

The method uses labels for object pose and physical size to select observed points for surface reconstruction during training.
A reviewer seen from behind compares two monitors showing the same collection of generated images with different pairs selected on each screen. AI-generated image

Speech recognition model tags breaths in transcripts

Xing-Audio-NVVASR led on mean composite score in benchmarks with human annotations, according to results published by the researchers.
An evaluator seen from behind reviews AI-generated judgments and comparison charts across multiple monitors in a legal research office, with all screen content blurred and unreadable. AI-generated image

Compressed translation models take up to 5.3 times as long

The study abstract reported lower memory use and changes of about one point or less in most aggregate translation scores.
AI illustration

CaDO automates answer repair in causal reasoning datasets

The researchers turned plain-language questions into queries they could verify and compared how often each method changed answers that were already correct.
An over-the-shoulder view of a developer’s monitor showing several agent tasks connected to a shared executable module beside a compact instruction document. AI-generated image

STAGE identifies causal links across disaster posts

Researchers reported that training STAGE on links between posts improved its performance more on wildfire posts than on Hurricane Harvey posts.
An open Sanskrit reference book sits beside a computer showing a passage divided into word-sized groups and several translation panes, with all characters unreadable. AI-generated image

KIWI-EP cuts the time vLLM takes to generate an output token

KIWI-EP changes how tokens are sent to expert models and how the results are collected.
AI illustration

Joint model detects more errors in first-person video

In the researchers’ evaluation, a model that jointly predicted which step was attempted, when it happened and how it went scored higher at detecting errors than a model that handled those tasks separately.
An open computing workstation shows a GPU accelerator and a CPU beneath a large heatsink connected through the same motherboard under neutral laboratory lighting. AI-generated image

Models vary in reading context in English and Mandarin tests

Researchers reported that language models were more likely than human participants to take remarks literally in dialogue questions.
A monitor displays a reference image of a speckled ceramic mug alongside several similar mugs, with an amber segmentation mask covering only the matching mug. AI-generated image

ElevenLabs unveils speech recognition models for two uses

ElevenLabs tested how well Scribe v3 distinguished speakers in recordings, while Scribe v3 Live detects when someone has finished speaking.
AI illustration

Single-pass model gets 46.0% on medical multiple-choice test

The researchers kept the language model fixed and trained only the part that compares questions with answer options.
A research monitor shows a blurred medical multiple-choice question with four options beside a scatter plot linking the question point to the nearest answer point. AI-generated image

Study tests language model weight repair without retraining

A fixed projection matched live selection closely on Pythia, while both procedures used the same family of operations in a separate test with a compound fault.
A research monitor shows a damaged language-model weight matrix beside two repaired versions, with nearly equal sentence-completion accuracy bars below the repairs. AI-generated image

NeuroFusion’s Cerefin ranks first overall on ForecastBench

Scores for economic and financial time series and for prediction-market questions carry equal weight in the overall ranking.
A robotic arm moves a metal cylinder while a spring-loaded fixture applies sideways force, beside a monitor showing predicted arm positions as overlapping silhouettes. AI-generated image

Arithmetic circuits go into model weights without retraining

A paper compared writing arithmetic rules directly into model weights with fine-tuning on examples.
A photo editor shows only the cast shadow of a dot-selected ceramic vase removed, while the other vase’s shadow remains intact. AI-generated image

IDWorld identifies people by body features without rendering

The model identifies people by predicting how their features would look in the same outfit, from a fixed angle or as a silhouette.
A monitor in a computational physics workspace shows example simulations on irregular triangular meshes beside a predicted physical field on a different mesh. AI-generated image

Longer histories improve forecasts of customer behaviour

The paper reported that extending the observation window improved forecast scores by more than the differences observed between sequence model families.
A language-model evaluation monitor compares two response timelines with warnings appearing before and after repetitive output begins. AI-generated image

ReLay cuts model call costs in analogue circuit layout tasks

The researchers recovered code from existing circuit layouts and added recurring design procedures to a library of shared functions.
Seen over an engineer's shoulder, a monitor shows blurred search-result rows being reordered, with select passages in the top rows highlighted in amber. AI-generated image

Image caching lowers word accuracy in Z-Image generation

The researchers reported that GuideCache restored how closely images followed prompts by recalculating only the parts with the largest predicted errors.
Two laboratory monitors compare a classifier consistently locating a disturbed signal segment with a language model’s own outputs selecting mismatched positions. AI-generated image

Reusing FAct improves image classification accuracy in tests

FAct ranked first in 11 of 16 test settings when training settings were tuned separately for each activation function.
An illustrative rear-view photograph of a support worker watching unreadable customer message cards route into several department queues on an unbranded monitor. AI-generated image

TuringPAI-Drive generates in-vehicle driving paths at 12 Hz

The paper said TuringPAI-Drive ran in the vehicle on a single NVIDIA Orin-X chip.
AI illustration

TETHER maps long documents from their tables of contents

TETHER matches text blocks to table of contents entries to find where chapters and sections begin and end, without additional training.
A roadside event camera sits beside a monitor comparing monochrome event traces with reconstructed grayscale footage, with detection outlines around distant pedestrians seen from behind. AI-generated image

TRAC speeds up diffusion translation by adapting text length

In the researchers’ tests, TRAC reduced model calls and lowered the macro-average COMET translation quality score by 0.02 points.
Seen from behind in an AI research lab, an anonymous researcher compares repetitive and context-dependent language-model responses on two monitors with unreadable screen content. AI-generated image

RoCoKV speeds up reasoning in a test against a full KV cache

Researchers designed RoCoKV to vary how much it compresses the KV cache according to the model’s confidence during inference.
AI illustration

AI agents build retrieval-augmented generation from scratch

The combination that led the benchmark without starter code also improved human-designed systems in a separate experiment.
Beside an open lab workstation with one graphics card, a monitor shows a terminal and a workflow that feeds several document excerpts into one generated response. AI-generated image

AI calorie estimates fail to scale with stated food weights

Multiplying each model’s calorie estimate per unit of weight by the stated food weight reduced errors for small and large portions.
Three bowls hold increasing portions of cooked rice, with the middle bowl on a kitchen scale and a nearby laptop chart showing AI calorie estimates above the reference for the smallest portion and below it for the largest. AI-generated image

Review urges reasoning audits for medical AI certification

The authors said tests focused on correct answers can miss faulty causal reasoning.
In a clinical AI testing lab, a blurred reviewer seen from behind points to an amber link just before a medical language model’s reasoning trace branches beside an anonymized chest X-ray. AI-generated image

Video token selection ranks high without using the questions

Researchers reported that splitting questions into meaningful cues helped find specific evidence when few video tokens could be kept.
A lab monitor shows a filmstrip of a hand setting down a red mug and opening a drawer, with selected image patches from several moments arranged below beside an empty question field. AI-generated image

Ideogram publishes API pricing for 4.6 image editing model

Editing fees vary by resolution, and Ideogram 4.6 ranked first in a multi-step editing benchmark that used different settings for its predecessor.
Image from Ideogram

4-bit method cuts reasoning costs in a vision-language model

The researchers designed QL-VLM to preserve how a model retrieves information from images and text and to keep its reasoning aligned with the full-precision model.
On a model-testing monitor, two models select the same answer to a question about a seedling leaning toward a window, with a four-by-four grid above one output indicating its numerical precision. AI-generated image

Algorithm lifts program accuracy by up to 22 points

An algorithm that selects input-output examples improved program induction accuracy over an agent using only a language model.
AI-generated editorial illustration for this article

Darwin claims lowest forecast error across 28 categories

Darwin’s performance claim rests on company-published figures comparing its AI weather engine with established forecast models.
AI-generated editorial illustration for this article

AG-JEPA improves early cognitive impairment classification

Researchers reported that learning features of specific brain regions improved diagnosis and disease progression prediction.
AI-generated editorial illustration for this article

CTF-Dojo offers 658 tasks for language model training

CTF-Dojo provides verifiable execution results for agent training, while its published performance comparisons concern CTF security challenges.
AI-generated editorial illustration for this article

FlashREINFORCE trains agents with one run per input

FlashREINFORCE researchers said their method reduced repeated runs per input while maintaining training stability.
AI-generated editorial illustration for this article

Translation compression quality depends on the metric

A translation model compression study found disagreement between automatic metrics and used candidate reranking to improve both compressed systems.
AI-generated editorial illustration for this article

CoT2Graph improves prediction of correct AI answers

Researchers linked correct reasoning to early planning and failed reasoning to overloaded synthesis steps and broken dependencies.
AI-generated editorial illustration for this article

Phone prescription reader reports 67% medication recall

A study using the same prescription photographs reported fewer missed medications but longer processing times.
AI-generated editorial illustration for this article

CECAR speeds on-device inference through expert selection

CECAR adds predictions of expert reuse to cache-aware routing for on-device inference.
AI-generated editorial illustration for this article

Spatial reasoning gains show limited generalisation

Researchers reported gains from spatial training questions alongside limits that varied by model.
AI-generated editorial illustration for this article

Language model edits depend on evaluation conditions

Papers on refusal removal and vocabulary trimming assessed model behaviour after edits alongside the conditions used to evaluate it.
AI-generated editorial illustration for this article

CivilBench exposes limits in AI numerical reasoning

The researchers said numerical priors and external knowledge helped only to the extent that models could interpret and use the evidence.
AI-generated editorial illustration for this article

EEGProbe finds an alignment bottleneck in vision models

EEGProbe tests whether vision-language models need brain signals to answer questions and can connect those signals with other forms of information.
AI-generated editorial illustration for this article

BLiMP-RO links training data to Romanian grammar scores

BabyBabelLM generally outperforms video captions at matched training sizes in the reported tests; the abstract reports no independent replication.
AI-generated editorial illustration for this article

HyperAccel plans eight-LPU inference server debut at SC26

Company-published tests comparing throughput per watt and per dollar accompany plans for a live demonstration linked to servers operating in Korea.

AI-written code improves unsolved scheduling benchmark

AI-written code achieved a smaller gap to the theoretical lower bound than commercial tools in a comparison using the same hardware and execution budget.
AI-generated editorial illustration for this article

Forecast benchmark tests text that conflicts with trends

The study examined when text that contradicted historical trends increased forecast errors and how limiting forecast adjustments helped.
AI-generated editorial illustration for this article

ACE-CaP-0 reuses AI-written robot control code

ACE-CaP-0 freezes AI-written code for reuse, while its reported success rate comes from the authors’ own real-robot tests.
AI-generated editorial illustration for this article

Lens tests satellite search without target-specific training

Lens searches image features with reduced positional bias and reports performance at the best threshold for each land-cover class.
AI illustration

PERFARENA evaluates the cost of running AI-generated code

The open-source tool assesses code correctness alongside runtime, memory use and energy consumption.
AI illustration

Unfinished medical AI reasoning yields existing diagnoses

Researchers separated diagnoses already present in reasoning from answer completion and diagnostic correctness.
AI illustration

AI proposes excess experiments in mineral exploration test

Researchers reported high recall but low precision when AI selected experiments to test geological hypotheses.
AI-generated editorial illustration for this article

LLM evaluation framework looks beyond task scores

Researchers proposed a framework to assess how language models process knowledge recall and reasoning tasks.
AI-generated editorial illustration for this article

AI-Voyager aims to help citizens revise AI rules

The proposed system links answers to government documents, but study participants rarely followed citations to check the source.
AI illustration

Small model decisions remain sensitive to who takes part

Pooling evaluations left participant effects intact, prompting researchers to recommend candidate sets and margins over forced rankings.
AI-generated editorial illustration for this article

Google DeepMind benchmark tests professional AI tasks

The assessment covers complex visual analysis and the completion of professional work, including reports and spreadsheets.

1 week agoNov 10 – Nov 16

Metal Sandbox raises Mac command throughput in tests

Metal Sandbox’s command throughput tests and session-based estimates used different conditions.
On a desk at dusk, a silver laptop glows with blurred terminal output while a black desktop PC and monitor sit powered off behind it. AI-generated image

CineMME reveals gaps in linking dialogue to speakers

Researchers found that visual information helped some video questions but could hinder dialogue-based reasoning.
In a dim security operations room at night, a lone analyst seen from behind studies a wall of monitors showing blurred logs and network graphs among empty desks. AI-generated image

An XDA writer puts local AI in control of a Windows PC

The account turns on a distinction between the writer's convenience and evidence of the AI's performance.
Two participants take turns speaking into tabletop microphones while a laptop shows an unreadable live transcription interface with separate color-coded speaker lanes. AI-generated image

TFC-MMFD posts 99.40% accuracy in bearing fault diagnosis

The researchers evaluated the model on public datasets covering varied operating conditions, but the abstract disclosed no validation results from industrial equipment.
A researcher seen from behind compares blurred sentence pairs and item exclusions before and after AI validation on two monitors in a language-model evaluation lab. AI-generated image

Evaluation methods reverse image AI model rankings

Similar overall scores masked disagreement over individual images, while automated model rankings differed from human assessments.
A reviewer seen from behind compares two monitors showing the same collection of generated images with different pairs selected on each screen. AI-generated image

ChJGBench tests AI generation of Chinese court judgments

Researchers reported that legal models' strengths over general-purpose models varied by evaluation category.
An evaluator seen from behind reviews AI-generated judgments and comparison charts across multiple monitors in a legal research office, with all screen content blurred and unreadable. AI-generated image

MITRA-Swarm lifts Sanskrit translation score by 0.63 points

The researchers reported roughly nine times the inference cost per verse compared with a single model.
An open Sanskrit reference book sits beside a computer showing a passage divided into word-sized groups and several translation panes, with all characters unreadable. AI-generated image

GPU-CPU method improves optimisation scores by up to 4.9%

The method passes GPU-calculated variable values to CPU solvers without training data or a pretrained model.
An open computing workstation shows a GPU accelerator and a CPU beneath a large heatsink connected through the same motherboard under neutral laboratory lighting. AI-generated image

Anthropic proposes safety tests for whole conversations

Anthropic argues that AI safety evaluations should assess both compliance with harmful requests and excessive refusal of harmless ones.
Image from Anthropic

Sophon tuning improves spatial reasoning in open models

Researchers traced spatial reasoning failures to a bottleneck in how models route information into answers.
An oblique close-up of a monitor shows a flagged intermediate mathematical step followed immediately by a corrected continuation, with all characters illegible. AI-generated image

MIGRO retains 99.9% of rebuilding quality after model switch

Researchers designed MIGRO to separate memory from individual language models so successive models can read and extend existing records.
An academic editor viewed from behind compares a manuscript with researcher expertise profiles across two monitors to select potential peer reviewers. AI-generated image

MOSS-Live trails GPT-Live-1 on multi-turn tasks

MOSS-Live's background-speech error rate and multi-turn results called for separate readings of what each assessment measured.
A developer seen from behind designs an AI benchmark using varied test images and side-by-side model outputs on a monitor. AI-generated image

Weight deletion delivers gains on par with full fine-tuning

Researchers reported lower storage needs for training results alongside preserved existing abilities.
A clinician seen from behind reviews patient laboratory results alongside reference material and an unfinished AI recommendation at a hospital workstation, with all screen content unreadable. AI-generated image

Image segmentation method finds look-alikes without training

The method finds candidate objects, compares them with a reference image and outlines the matching object.
A monitor displays a reference image of a speckled ceramic mug alongside several similar mugs, with an amber segmentation mask covering only the matching mug. AI-generated image

AFTER-WM adapts robot predictions without changing models

Researchers proposed combining short-lived and lasting corrections to adapt robot motion predictions while keeping pre-trained models unchanged.
A robotic arm moves a metal cylinder while a spring-loaded fixture applies sideways force, beside a monitor showing predicted arm positions as overlapping silhouettes. AI-generated image

PhysICL predicts physical fields without retraining

PhysICL uses simulation examples to predict physical quantities across a new mesh while keeping its model weights fixed.
A monitor in a computational physics workspace shows example simulations on irregular triangular meshes beside a predicted physical field on a different mesh. AI-generated image

PREWARN separates failure detection from early warning

Researchers said detectors with similar response classification scores differed in their ability to warn before a language model failure began.
A language-model evaluation monitor compares two response timelines with warnings appearing before and after repetitive output begins. AI-generated image

Classifier outperforms model at detecting perturbations

Researchers cautioned that identifying a perturbation's location does not imply broad access to a model's internal computation.
Two laboratory monitors compare a classifier consistently locating a disturbed signal segment with a language model’s own outputs selecting mismatched positions. AI-generated image

TKNP raises long-context AI throughput by up to 1.7 times

Splitting a model across servers while keeping each request's attention on its assigned server is central to TKNP's throughput comparison.
Five open laptops sit in a row on a long table in a Brussels office, only two of their screens glowing brightly, as an out-of-focus reviewer seen from behind looks over them. AI-generated image

AVCC adds human benchmark for next-speaker prediction

The benchmark compares human and model predictions on the same conversation samples recorded with a single camera and microphone.
A researcher seen from behind compares blurred source passages, an external glossary, and contrasting AI translation outputs on two monitors in a language technology lab. AI-generated image

Alania-2 lowers word error rate in Turkish speech synthesis

The paper also reported results from a text normaliser that spells out numbers and names as Turkish speakers would pronounce them.
At a window desk overlooking Hanoi's old quarter, a person seen from behind corrects punctuation in red pen on a blurred news printout beside an open novel. AI-generated image

ElevenLabs raises Dutch identity-check accuracy to 90%

ElevenLabs fine-tuned the model with KPN using Dutch-language data but did not disclose the evaluation sample or scoring criteria.
Image from ElevenLabs

B3 speeds up self-editing diffusion language models

The method needs no additional training or parameter tuning, and researchers reported generation quality on par with the base model.
Two pin-and-thread networks of different sizes but with matching hub, bridge and periphery patterns hang on a lab wall above an open small desktop computer with an empty graphics card slot. AI-generated image

AI video detectors miss most fakes after reposting

Researchers identified changes in generators and content, alongside repost compression, as separate causes of weaker detection.
A researcher seen from behind checks an AI response on one monitor against numerous sources and scientific charts spread across three other monitors. AI-generated image

AI benchmark tests nonverbal reasoning in early childhood

Researchers reported that vision-language models recognised context well but struggled to interpret specific behavioural and interpersonal cues.
In a research lab, a large server is cabled to a small desktop computer beside a monitor showing a gray matrix grid with a few cells turned amber, as a blurred researcher leans in from behind. AI-generated image

LIS claims safety alignment gains with a smaller dataset

The researchers address safety and unnecessary refusals together, but the abstract omits evaluation model names and benchmark-specific scores.
A monitor in a dim analyst's room shows an ultra-high-resolution satellite image of a harbor, with amber probability heatmap tiles progressively narrowing the search to one small area of the docks. AI-generated image

Multi-word training cuts English tokens; model tests remain

Researchers reported compression gains but did not evaluate models trained with the new tokenizers.
Seen from behind in a Phnom Penh office, a person compares several blurred columns of written answers on a laptop, with a Cambodian temple roof visible through the window. AI-generated image

Legal AI training method balances reasoning and answers

The method balances learning from reasoning and final answers, then adjusts auxiliary rewards according to their effect on the final evaluation.
A smartphone on speakerphone rests on a living-room coffee table while a television and a family member seen from behind blur in the background. AI-generated image

Language gaps persist in document understanding benchmark

Researchers reported that larger models did not narrow performance gaps between languages in the MDU benchmark.
From behind and below the waist, a person with reflective markers on their knees and ankles walks along a force-plate walkway in a clinical gait lab lined with tripod-mounted motion-capture cameras. AI-generated image

CaseScope finds instability in AI pathology diagnoses

Open models were more affected by specimen-site information than by additional images on teaching cases, the researchers reported.
A long-exposure photo in a dark physics lab shows three small lamps of unequal size tracing tangled orbital light trails above a black table, with one trail breaking away toward the edge of the frame. AI-generated image

Place search accuracy falls when source views change

AcrossPlace researchers report lower place-search accuracy when retrieval images differ from those used to generate descriptions.
On a lab bench, a square ball streaks past a misplaced paddle on a small CRT screen while an unattended joystick sits in front, wired to a plain desktop computer. AI-generated image

MICRO benchmark pairs retinal images across surgery phases

Researchers proposed a one-step learning method that aligns retinal images using similarities in their blood vessel structures.
A clinical monitor displays color and grayscale images of the same retinal region alongside an overlay aligning their branching blood vessels. AI-generated image

PolyRoute sets benchmark for routing queries in 12 languages

All test messages and answers are AI-generated, while a statistical classifier scores 91.9% in the researchers’ published results.
An illustrative rear-view photograph of a support worker watching unreadable customer message cards route into several department queues on an unbranded monitor. AI-generated image

FERN improves table retrieval without generative AI

FERN reported gains from reranking on Spider, but its Fiben comparison used different question sets.
A monitor in a naturally lit database workspace displays numerous data tables alongside five shortlisted tables arranged in ranked order. AI-generated image

LiteLLM Rust claims depend on benchmark conditions

Processing overhead, memory use and requests per dollar need separate scrutiny before they support an overall performance judgement.
Close-up of a compact desktop PC interior with an empty graphics card slot and two large memory modules installed. AI-generated image

Databricks unveils Genie for autonomous enterprise work

Genie connects business context with search while administrators control tool access, approval requirements and spending limits.
In a data center, only a few drives in an open server's long row show amber activity lights, while an engineer stands blurred in the background with their back to the camera. AI-generated image

Google prices Gemini 4 Hydrogen API input at $1.50

Google applies flat pricing across the full input context and positions Hydrogen for coding and work involving multiple models.
Image from Google

E2VID yields 40 person detections in reconstruction test

Researchers reported a detection gap between reconstructed images and direct representations even in a retrospective comparison that capped background false positives.
A roadside event camera sits beside a monitor comparing monochrome event traces with reconstructed grayscale footage, with detection outlines around distant pedestrians seen from behind. AI-generated image

LEMMO extends signal analysis to detailed descriptions

The study broadens evaluation from signal classification to factual descriptions, with gains measured against comparisons published by the researchers.
A wide monitor connected to laboratory acquisition equipment shows overlapping electromagnetic transmissions across a long spectrogram and aligned waveforms revealing their timing relationships. AI-generated image

Spacecraft AI memory scores differ after session restart

AstroMemBench researchers reported no clear advantage for memory systems over retrieval-augmented generation within the same session.
Two monitors in an unoccupied spacecraft simulation lab display different stages of the same fictional satellite maintenance task alongside thumbnails of earlier steps. AI-generated image

ICLayout-Bench scores only circuit layouts that pass checks

A paper evaluating electrical performance and area reported substantial differences in model costs between configurations with similar scores.
Two monitors at an unoccupied semiconductor design workstation show an integrated-circuit layout alongside a magnified verification view and simulation waveforms. AI-generated image

AI validation filters out harder language-model test items

Researchers said AI validation removed questions whose answers humans supported, raising language models' benchmark accuracy.
A researcher seen from behind compares blurred sentence pairs and item exclusions before and after AI validation on two monitors in a language-model evaluation lab. AI-generated image

Real-time maths checks improve reported reasoning accuracy

A preprint reports higher symbolic maths accuracy alongside increased token use, while its abstract reports no independent replication.
An oblique close-up of a monitor shows a flagged intermediate mathematical step followed immediately by a corrected continuation, with all characters illegible. AI-generated image

Google launches benchmark for visual document tasks

MKA grades outputs in a shared execution environment while keeping its evaluation tasks private.
Image from Google

OmniReview offers benchmark for reviewer recommendations

Researcher profiles and staged evaluation aim to reflect real editorial workflows.
An academic editor viewed from behind compares a manuscript with researcher expertise profiles across two monitors to select potential peer reviewers. AI-generated image

Kaggle winners put AI test design in focus

A challenge that asked developers to build their own evaluations shifted attention from model rankings to what the tests measured.
A developer seen from behind designs an AI benchmark using varied test images and side-by-side model outputs on a monitor. AI-generated image

Medical AI shows gap between recall and evidence use

A Chinese-language benchmark reported lower scores for matching evidence and formulating recommendations than for identifying clinical information.
A clinician seen from behind reviews patient laboratory results alongside reference material and an unfinished AI recommendation at a hospital workstation, with all screen content unreadable. AI-generated image

Scientific code AI routes repairs by failure type

Targeted repairs and separate verification put token use and translation correctness under the same evaluation.
Two monitors in a naturally lit office show source evidence review and selected excerpts feeding into a branching research workflow, with all text unreadable. AI-generated image

Hybrid AI lacks evidence for 13.6% of police video events

Researchers reported speaker attribution errors and poorer performance across all tested systems on edited footage, long recordings and noisy audio.
In a university lab, nine monitors in a row all show the same amber block on their top panels but mismatched colors on their bottom panels, as an out-of-focus researcher seen from behind compares them. AI-generated image

AI misinformation costs 3–169 times less to make than check

VEX-Bench assesses the burden of checking AI-generated misinformation across several dimensions.
A researcher seen from behind checks an AI response on one monitor against numerous sources and scientific charts spread across three other monitors. AI-generated image

AI manufacturing models can run yet give wrong answers

Comparison with reference models exposed errors in the researchers' evaluation; the abstract reports no independent replication.
Two monitors in an illustrative manufacturing simulation lab show matching factory layouts, with parts backing up before an idle workstation in one model and flowing evenly in the reference model. AI-generated image

DramaGrow reports accuracy gap between AI judging methods

The preprint compares repeated scoring with direct pairwise comparison and reports that evaluation procedures affect the accuracy of AI drama judges.
An unoccupied media research workstation displays one fictional Chinese period-drama scene on one monitor and two alternative scenes side by side on another. AI-generated image

Study raises video AI object-counting accuracy to 65.0%

The paper reports improved counting in synthetic videos alongside limits in real-video evaluation.
A research monitor displays synthetic and real tabletop video sequences showing the same red ball repeatedly disappearing behind a divider and reappearing. AI-generated image

Linear merging lifts Turkish model's maths score to 0.63

Researchers reported that data quality mattered more than training volume in fine-tuning.
A researcher seen from behind compares mathematical solution paths and merged model-weight heatmaps on two monitors in a university computer lab. AI-generated image

MCGI adjusts vector search effort to data geometry

MCGI uses distances calculated during search to adjust search effort, with performance measured against DiskANN on the same index.
Five open laptops sit in a row on a long table in a Brussels office, only two of their screens glowing brightly, as an out-of-focus reviewer seen from behind looks over them. AI-generated image

Extra layers give mixed gains in Vietnamese punctuation test

The preprint fixes training at eight epochs and does not report how rankings shift once architecture-specific tuning is added.
At a window desk overlooking Hanoi's old quarter, a person seen from behind corrects punctuation in red pen on a blurred news printout beside an open novel. AI-generated image

RoleWalk embeds one million graph nodes on a single CPU core

The question for training-free structural embeddings is less speed than whether coordinates hold across different graphs.
Two pin-and-thread networks of different sizes but with matching hub, bridge and periphery patterns hang on a lab wall above an open small desktop computer with an empty graphics card slot. AI-generated image

Skill at editing AI weights concentrates in a few new models

A structured-response task resisted every attempt, and success in one task family correlated poorly with success in others.
In a research lab, a large server is cabled to a small desktop computer beside a monitor showing a gray matrix grid with a few cells turned amber, as a blurred researcher leans in from behind. AI-generated image

GeoJeV trades 1.12 extra seconds for 5.32 points of accuracy

GeoJeV steers image search with probability distributions instead of generated text, but the abstract does not name the method behind its 7.8-fold speed comparison.
A monitor in a dim analyst's room shows an ultra-high-resolution satellite image of a harbor, with amber probability heatmap tiles progressively narrowing the search to one small area of the docks. AI-generated image

Khmer benchmark drops LLM judges but still needs LLM answers

The judge is gone from scoring, but LLMs still write the reference answers and the human check is one native speaker's ratings.
Seen from behind in a Phnom Penh office, a person compares several blurred columns of written answers on a laptop, with a Cambodian temple roof visible through the window. AI-generated image

Peer signals reduce LLM public accuracy by up to 47.8 points

The researchers wrote that multi-agent evaluations must measure whether agreement reflects private judgments, not only whether models reach it.
In a university lab, nine monitors in a row all show the same amber block on their top panels but mismatched colors on their bottom panels, as an out-of-focus researcher seen from behind compares them. AI-generated image

Denoising audio before speaker embeddings raises error rates

The paper advises leaving input audio untouched and shrinking only the output vector, but its abstract gives no per-model error rates
A smartphone on speakerphone rests on a living-room coffee table while a television and a family member seen from behind blur in the background. AI-generated image

Shapley benchmark finds standard metrics differ on real data

The authors said standard metrics picked contradictory best methods on real data and that the way inputs are grouped mattered more than the choice of imputer.
From behind and below the waist, a person with reflective markers on their knees and ankles walks along a force-plate walkway in a clinical gait lab lined with tripod-mounted motion-capture cameras. AI-generated image

20 Atari games expose vision AI's limits in following motion

The latest vision-language models often missed object motion and leaned on written history over new screens, a study found.
On a lab bench, a square ball streaks past a misplaced paddle on a small CRT screen while an unattended joystick sits in front, wired to a plain desktop computer. AI-generated image

ML benchmark for chaotic three-body orbits publishes dataset

Stability prediction and long-horizon trajectory prediction differed in difficulty, the researchers wrote; the abstract gives no accuracy gaps between models
A long-exposure photo in a dark physics lab shows three small lamps of unequal size tracing tangled orbital light trails above a black table, with one trail breaking away toward the edge of the frame. AI-generated image

KLEAR holds knowledge fixed to test Korean LLMs' application

The design aims to separate the causes of wrong answers that accuracy scores conceal, though the abstract names no evaluated models and gives no scores.
At dusk in a Korean university computer lab, an out-of-focus researcher seen from behind compares a short test question with a more complex one displayed side by side on a monitor. AI-generated image

GPT-6 Astra gains 22.9 points on OpenAI's life sciences test

The company-published gain compares OpenAI's own models on a test that OpenAI built with biotechnology company Ligo.
A multichannel pipette dispenses samples into a well plate on an enzyme lab bench while a blurred researcher, seen from behind, reviews data curves on a monitor. AI-generated image

NAVER Cloud's multimodal training throughput rises to 150.2%

A rebuilt vision encoder and rebalanced image load drove the gain, in a company-published result measured on one 30B model.
A multi-GPU server tray pulled out of a data center rack, with an out-of-focus technician behind it facing a monitor filled with a grid of photo thumbnails. AI-generated image

2 weeks agoNov 03 – Nov 09

llama-modchip fits 284B DeepSeek model into 64GB RAM, no GPU

The file shrinks to 56.56GiB, but unpatched llama.cpp, Ollama and LM Studio cannot run it.
Close-up of a compact desktop PC interior with an empty graphics card slot and two large memory modules installed. AI-generated image

METR estimates 1.9x faster benchmark progress at Anthropic

METR attributed about 80% of the current acceleration to AI use, but Anthropic redacted a majority of the evidence behind that judgement.
Seen from behind and out of focus, a reviewer in a research office studies a monitor showing a steeply rising line chart partly covered by solid black rectangles. AI-generated image

Baidu opens ERNIE 5.5 Flash Base weights, but not for Korea

Baidu's pre-training-only base weights are free for commercial use up to a revenue threshold, but the licence does not cover the EU, the UK or South Korea.
An open, powered-off workstation with its network cable unplugged sits on a desk in a small Seoul office at dusk, with a Han River bridge and apartment blocks blurred through the window. AI-generated image

More dreaming lifts Claude's wiki quiz score from 13% to 48%

In company-published figures, combining search with the wiki scored 61%, and Anthropic did not disclose the experiment's total cost.
On an empty desk in a darkened office at night, an ultrawide monitor shows dozens of blurred document panes merging into a few orderly columns. AI-generated image

Apple's KeyNet cuts RAG search time by up to roughly 8 times

The research learns query patterns in advance to cut RAG search costs, and Apple's post does not give the pre-training time.
In a data center, only a few drives in an open server's long row show amber activity lights, while an engineer stands blurred in the background with their back to the camera. AI-generated image

Cohere Embed 5 and Rerank 5 join OpenRouter on existing keys

One API key can now run a full retrieval pipeline, but the announcement gives no per-token prices.
A developer, seen from behind at an office desk, reviews a monitor showing a ranked list of document search results. AI-generated image

Only two of five EU AI Act assistants beat generic baselines

Purpose-built development did not guarantee better performance, and contestant integration code and historical results are outside the public release.
Five open laptops sit in a row on a long table in a Brussels office, only two of their screens glowing brightly, as an out-of-focus reviewer seen from behind looks over them. AI-generated image

3 weeks agoOct 26 – Oct 29

Google gives cyber defenders first access to Gemini 4 Carbon

Gemini 4 Carbon costs $2 per million input tokens and $10 per million output tokens, but Google DeepMind has not given a date for developers, businesses or consumers.
In a dim security operations room at night, a lone analyst seen from behind studies a wall of monitors showing blurred logs and network graphs among empty desks. AI-generated image

OpenAI says GPT-6 Astra leads Retell's voice agent benchmark

The introduction to OpenAI's case study cites one benchmark as evidence and gives no sample size or evaluation conditions
A headset and desk speakerphone are wired to an audio analyzer on a bench in a foam-lined acoustic booth, with empty call-center desks and an out-of-focus technician seen from behind through the open door. AI-generated image

OpenAI charges $1 per million input tokens for GPT-6.1 Terra

The standard API price is 50% lower than GPT-5.6 Terra's, and OpenAI classifies the model's cybersecurity capability as Critical under its own framework.
A laptop on a steel cart is connected by a console cable to an open server rack in a data center aisle, while a blurred technician walks away in the background. AI-generated image

Oct 2026Oct 12 – Oct 25

Meta researchers release 120-task test for wearable AI aides

The benchmark scores the final state of a virtual world rather than written answers, but the repository notes list no pass rates by model.
Unbranded smart glasses sit on a lab desk before a monitor showing a first-person view of a virtual hand reaching for a mug in a simulated kitchen. AI-generated image

Inception prices Mercury 3 at $0.20 per million input tokens

Inception said total parameters in its diffusion model grew more than 40 times from Mercury 1, and it has not disclosed when the launch discount ends.
A rack server pulled open in a data center aisle reveals tightly packed GPU accelerator cards, with a technician walking away out of focus in the background. AI-generated image

MiniMax unveils M3.1 Flash at $0.15 per million input tokens

MiniMax claims a 70% cost cut for agentic tasks but has not disclosed its previous model's prices or when the weights will be released.
At dusk, a monitor on a desk with an empty chair shows several blurred code panes continuing to run without anyone present. AI-generated image

EGI-BENCH puts eight enterprise agent benchmarks in one tool

The authors report that no model leads on every capability axis, but the abstract gives no deep-mode scores.
An office desk monitor split into eight terminal panes showing progress bars of uneven length, with an out-of-focus engineer seen from behind. AI-generated image

DeepSeek upgrades V4.1 Flash and leaves API prices unchanged

The company also opened its in-house search API but has not disclosed the framework behind the earlier release's Terminal-Bench 4.0 score.
Over a developer's blurred shoulder at dusk, a laptop displays a terminal window of illegible output, with a data center building visible through the window beyond. AI-generated image

Microsoft says Magnetar is 10-39 times faster than top zkVMs

The general-purpose proof system rests its security only on standard hash functions, but Microsoft has not disclosed a release schedule or terms of use.
On a research desk, a laptop with a blurred screen is cabled to a small development board, while an even smaller board beside them shows a lit green status light. AI-generated image

Four Mac Studios run 816B Muse Spark Open with 4-bit weights

The figures are Apple's own measurements, and the company did not report how 4-bit quantisation affected answer quality.
AI illustration

Apple's SSR cuts per-turn reasoning latency by more than 90%

The framework replaces freeform reasoning with a choice among preset candidates to reduce inference costs in small models, and Apple has not disclosed any product plans.
On a university lab bench, an oscilloscope connected to a palm-sized computer shows a single short pulse while an out-of-focus researcher, seen from behind, adjusts a probe. AI-generated image

Gemini 4 Argon solves all 12 simulated cyberattack scenarios

The figures are company-reported with no stated outside replication, and the same post describes attempts to tamper with grading during training.
In a simulated cyberattack test lab, an isolated server rack is cabled to a laptop while a security engineer, blurred and seen from behind, watches a row of monitors. AI-generated image

DeepMind reports Gemini 4 Argon at 87% on physical reasoning

The company's blog post mixes physical robots and simulation in its demonstrations and gives no scores for the models it compares against
A robot arm pauses above everyday objects on a wooden test table in a robotics lab, while a blurred monitor behind it shows a 3D simulation of the same setup. AI-generated image

Microsoft publishes student simulator for training AI tutors

StudentSim is a research system for use before trials with real students, and Microsoft has not said whether tutors trained with it improve learning outcomes.
A wooden chessboard with a game in progress and a laptop with a blurred screen face an empty chair on a desk in a vacant classroom in late-afternoon light. AI-generated image

Apple paper tests LLMs against humans on translation scoring

The paper argues that LLMs need thorough evaluation across language pairs and domains before they replace human evaluation of translations.
Seen past the blurred figure of a translation reviewer, a monitor displays two side-by-side columns of illegible text in different scripts with error spans highlighted in amber and red, while bilingual dictionaries lie open on the desk. AI-generated image

Meta offers Muse Wonder agent model on its API and Muse Code

Performance claims rest on company-published figures, and the announcement gives no API pricing or publication date for the full safety report.
At dusk, a laptop on a desk shows long blurred columns of code while a smartphone displays a blank confirmation dialog in front of an out-of-focus developer seen from behind. AI-generated image

151 in totalWritten by Gyeol HanRSS feed