groundy

Groundy — independent coverage of developer tools, infrastructure, and platforms

industry

DeepSeek V4 Cache Discounts, Not Peak-Valley Pricing, Shape Cost Decisions

DeepSeek V4 has no peak-valley pricing. The real lever is a fiftyfold cache discount and hard concurrency caps that force teams to optimize for prefix reuse before July 24.

6min·10:32·jul 06
devtools

Vite+ Beta: MIT-Licensed Now, Paid Tier Later

Vite+ beta ships as MIT-licensed open source with unified toolchain commands and enterprise templates, but VoidZero deferred commercial tier pricing until the 1.0 release.

6min·15:58·jul 06
policy

Stochastic Dominance Reveals Where RLHF Safety Filters Hide Tail Risk

Standard Safe RLHF optimizes for average harm reduction, leaving rare catastrophic outputs hidden in the tail. Stochastic dominance forces teams to bound the entire harm.

devtools

Vercel's Agentic Infrastructure Push Outpaces Pricing Transparency

Vercel is building isolated execution environments for agent workloads, but without published capacity or pricing, teams cannot compare the platform against AWS Fargate or.

oss

Box3D Launches as Open-Source 3D Physics Engine

Box2D author Erin Catto released Box3D, a C17 open-source 3D physics engine backed by his day job at Kintsugiyama, giving game developers a credible alternative to Nvidia's.

models

VLA Grounder Tests Language Conditioning to Optimize Black-Box Vision-Action Models

VLA Grounder shows frozen vision-language-action models improve when language is an optimizable input. Success rates increased from 12.6% to 38.7% for pi0 and 24.4% to 63.9%.

devtools

Cursor iOS Privacy Migration Shows Why Mobile IDEs Can't Be Audited

Cursor's iOS app migrated users to a new privacy mode without consent, exposing how iOS sandbox design prevents developers from auditing or reversing what mobile IDEs do with.

models

Black-Box LLM Architecture Inference: What API Restrictions Reveal About Hidden Model Structure

NightVision recovers transformer hidden dimension to within 23% error using only single logprob output and timing, proving API restrictions meant to protect model IP instead.

featurededitor's picks
most popularlast 6 days · by traffic
tophighest rated
feedall 729 articles · reverse chronological
  1. jul 06industryDeepSeek V4 Cache Discounts, Not Peak-Valley Pricing, Shape Cost Decisions
  2. jul 06devtoolsVite+ Beta: MIT-Licensed Now, Paid Tier Later
  3. jul 06policyStochastic Dominance Reveals Where RLHF Safety Filters Hide Tail Risk
  4. jul 06devtoolsVercel's Agentic Infrastructure Push Outpaces Pricing Transparency
  5. jul 06ossBox3D Launches as Open-Source 3D Physics Engine
  6. jul 06modelsVLA Grounder Tests Language Conditioning to Optimize Black-Box Vision-Action Models
  7. jul 06devtoolsCursor iOS Privacy Migration Shows Why Mobile IDEs Can't Be Audited
  8. jul 06modelsBlack-Box LLM Architecture Inference: What API Restrictions Reveal About Hidden Model Structure
  9. jul 06modelsInduceKV Tests Continual Learning for Multimodal LLMs Without Expanding Cache
  10. jul 06cultureUnit Labor Costs Hit Post-War High as Productivity Decouples From Wages
  11. jul 05cultureJune 2026 Labor Force Contraction Tests Structural Detachment Thesis
  12. jul 05devtoolsTab Completion Hides a Vigilance Drop That Copilot Metrics Miss
  13. jul 05cultureJune's Jobs Polarization Reveals AI-Era Skills Repricing
  14. jul 05agentsBOUNDARY_SYNC: Why Multi-Agent Representational Coupling Is the New Coordination Failure Mode
  15. jul 05devtoolsKimi K2.7 Code Lands in GitHub Copilot: What the Integration Excludes
  16. jul 02modelsSonnet 5 vs GPT-5.5: Pricing, Benchmarks, and the Switching Math
  17. jun 29agentsDo Multi-Agent RAG Systems Write Better READMEs Than One Agent?
  18. jun 29securityJailbreaks Hidden in Image Pixels Slip Past Editors' Text Guardrails via an Empty Prompt
  19. jun 29infraDoubao 2.1 Pro: What 180 Trillion Daily Tokens Means for Inference Infrastructure
  20. jun 29devtoolsVercel Firewall in the CLI: What's Still Missing
  21. jun 29infraEvery CUDA Kernel Pays a Launch Tax: The Host-to-Device Walkthrough
  22. jun 29modelsHow LLMs Fuse Conflicting Facts: Single-Source vs Multi-Source Truth
  23. jun 28securityLinux Foundation Akrites Centralizes Open-Source Vulnerability Disclosure
  24. jun 28modelsLinear Transformers Get a Learnable Kernel: Does Flexformer Change the Efficiency Tradeoff?
  25. jun 28securityWhy LLM Prompt Injection Persists: Instructions and Data Share Embeddings
  26. jun 28cultureGenerative AI Moves the Freelance Bottleneck From Tasks to Skill Repricing
  27. jun 28policyUncertainty-Aware Reward Discounting Cuts Reward Hacking 93.6% in a Preprint
  28. jun 28securityWhen Bots and Agents Post CVEs in PRs, Reporters Inherit the Triage Burden
  29. jun 28securityRuntime vs Build-Time SBOMs: Why Your Container Runs Uncatalogued Code
  30. jun 28industryElkjøp's Next.js Move Shows Vercel Wants Retail Operations, Not Just Websites
  31. jun 28agentsAgentic AI Turns Location Trails Into a Re-Identification Tool
  32. jun 28modelsHuawei Ships CUDA-Free AI Compute On-Device, but Ascend Quantization Accuracy Is Unverified
  33. jun 28infraVercel Montreal Region: Audit Residency Before You Migrate
  34. jun 28agentsHow a Human-Agent Team Lifts One Video Into 4D Interactions
  35. jun 28ossSafetensors vs Pickle: Why Hugging Face Chose It After the Security Audit
  36. jun 28modelsDo Multimodal RAG Models Ignore Late Evidence? A Primacy Bias Test
  37. jun 28securityOpenAI's Agent Link Safety Isolates the Fetch, Not Prompt Injection
  38. jun 28agentsCan LLM Agents Learn Cooperation Laws From Embodied Play?
  39. jun 28securityNo Verified 'React2Shell' Bulletin Exists: What Next.js Teams Should Check
  40. jun 28modelsCan Deep Learning Design RF Power Amplifiers Without Full EM Simulation?
  41. jun 28securityVercel on the Axios npm Compromise: Platform Scanning Has a Blind Spot
  42. jun 28agentsGovern the Repo, Not the Agent: A New Risk Metric for AI-Native Code
  43. jun 28cultureLLM-Generated VeriFast Specs Shift the Trust Bottleneck from Proofs to Review
  44. jun 28infraGLM-5.2 on vLLM and Ascend: Open Weights Beyond NVIDIA
  45. jun 28ossHugging Face Is Absorbing Computer Vision Into Vision-Language Models
  46. jun 27agentsCan an AI Agent Catch Cryptographic Misuse Before It Ships? Chai Tests the Claim
  47. jun 27devtoolsVercel's CLI Is a Deployment Path, Not a Control Plane
  48. jun 27infraHow Vercel Runs Its Own CDN in Front of Discourse: A Self-Dogfooding Case Study
  49. jun 27industryByteDance's Doubao Seed 2.1 Pro: Production-Grade Claims, Vendor-Graded Evidence
  50. jun 27devtoolsGLM-5.2 Goes Open Weights: What the Long-Horizon Coding Pitch Leaves Out
load older →