News & Releases
[macro] creeta 3-axis IA
langchain-perplexity 1.3.1: Routing Logic, SSE Correction, Explained
1.3.0 added use_responses_api for Perplexity's Responses endpoint; 1.3.1 removed the SSE shim 0.34.0 required.
vLLM v0.22.0: DeepSeek V4, Rust Frontend, Concurrent rc0 Explained
459 commits, a dedicated DeepSeek V4 package, Rust frontend, and an rc0 that's one CI fix. What matters and what doesn't.
OpenAI Rosalind Biodefense Program: Criteria, Partners, and Caveats
OpenAI extends GPT-Rosalind to vetted public health orgs — free, sponsored, and still gated from general use.
Google Beam Goes Multi-Person: How the 3D Group Call Experiment Works
Google I/O 2026 extended Beam to multi-person calls. Here's the AI pipeline, the $24,999 display, and where the gaps are.
Gemini for Science: What Google I/O 2026 Introduced for Researchers
Google's I/O 2026 AI research suite: literature triage, hypothesis tournaments, and ERA outperforming CDC forecasts.
China's AI Token Futures Plan: What Shanghai Is Building in 2026
Shanghai Futures Exchange is prototyping AI token futures — forward contracts on LLM consumption costs. Here's the technical picture.
DiffusionBlocks: Sakana AI's Block-Wise Training for ICLR 2026
DiffusionBlocks trains one residual block per step, reducing activation memory B× with competitive or better accuracy.
Robinhood MCP: How AI Agents Now Trade Stocks and Make Purchases
Robinhood opened its brokerage and card infrastructure to MCP-compatible AI agents. Here's what the implementation looks like technically.
langchain-perplexity 1.3.0: ChatPerplexity Gets use_responses_api
ChatPerplexity gains use_responses_api in 1.3.0: auto-routes to Perplexity's Agent API for real-time search.
Anthropic Mid-Run Constraint Blocks: Cache-Safe Prompt Updates 2026
Mid-conversation constraint injection in v0.105.0 preserves prompt cache continuity across long inference runs.
Microsoft Copilot Cowork: Prompt Injection Exfiltrates M365 Files
A 5-line poisoned Skills script silently exfiltrates SharePoint data via Copilot Cowork — no approval gate, no CVE, no patch.
Project Genie + Street View: Real-World Simulation Lands in Genie 3
Genie 3 generates interactive worlds from real Street View geometry. Waymo is already using it for rare-event training.
DeepMind's Running Guide Agent: On-Device Gemma 4 for Blind Athletes
DeepMind's chest-mounted AI system lets blind runners navigate independently using dual-path on-device inference—no cloud, no tether.
Anthropic SDK 0.105.1 and 0.105.2: PyPI Trusted Publishing Hotfix
Two rapid patches followed Anthropic's 0.105.0 drop. Here's what broke, why, and which version to pin.
Claude Code May 2026: Permission Fixes, /code-review --fix, MCP Auth
Seven builds in one week: four Bash/PowerShell sandbox bugs patched, /code-review --fix lands auto-apply, and a serious MCP auth credential leak is closed.
Gemini for Science at I/O 2026: How Each Research Tool Works
Three experimental AI research tools launched at I/O 2026. What Literature Insights, Co-Scientist, and AlphaEvolve each actually do.
Google Workspace at I/O 2026: Docs Live, Gmail Live, and Gemini Spark
Docs Live, Gmail Live, Gemini Spark, Sheets one-shot: I/O 2026 Workspace features and who gets access first.
Anthropic 0.105.0 Under the Hood: Output Attribution and File Caps
v0.105.0 adds granular output-type attribution and configurable upload caps—here's what they do and when to use them.
vLLM v0.21.0 Production Update: KV Offload and Multi-Server Port Bug
v0.22.0 doesn't exist yet. v0.21.0 ships KV offload, spec decode, and a multi-server port bug still under review.
Claude Code v2.1.144–v2.1.154: What Shipped in Nine Days
Ten patches in nine days: pinned sessions, four security fixes, /code-review --fix, and skill-level tool gating.
SuperGrok Subscription Now Unlocks grok-build-0.1 in Kilo Code
SuperGrok and X Premium+ subscribers can now authenticate into Kilo Code and run grok-build-0.1 inside VS Code or JetBrains — no API key management required.
Codex CLI 0.134.0 and 0.135.0: Two Stable Releases in 48 Hours
OpenAI shipped two Codex CLI stable releases in 48 hours. What changed, what broke, and why the cadence matters.
Anthropic Python SDK 0.105: Opus 4.8 and Mid-Session System Prompts
Three SDK releases in 7.5 hours ship claude-opus-4-8 support, mid-conversation system blocks, and finer output usage reporting.
xAI grok-build-0.1 API Public Beta: Token Costs and SDK Support
xAI's coding model exits the $299 CLI gate. Here's what the public API beta actually offers developers.
Grok Build Lands in OpenCode and Kilo Code: xAI's 13-Day Rollout
xAI shipped grok-build-0.1 to three developer tools in 13 days. Here's what each integration covers and how to pick the right surface.
Codex CLI 0.135.0: Doctor, Vim Text Objects, and the 0.136 Alpha
OpenAI's 0.135.0 stable is a diagnostics and polish cycle. What moved in the TUI, Vim mode, and remote transport.
Command A+ 2026: Benchmark Results, Citation Tags, and Enterprise Fit
Cohere's first open-weight frontier model: benchmark gaps, native citation design, and the enterprise sovereignty case.
grok-build-0.1: Model Spec, Pricing, and the Beta Rollout Story
xAI's grok-build-0.1 hit public beta in May 2026. Here's what the spec says — and what the caching incident revealed.
Claude Opus 4.8: Coding Benchmarks and Agentic Upgrades 2026
Anthropic ships Opus 4.8 with 69.2% SWE-Bench Pro, mid-conversation system messages, and adaptive thinking.
Codex CLI 0.135.0-alpha.2: Scope, Diff, and the Release Notes Error
Two alpha releases in three hours, 529 files changed. Here's what the diff says when the release notes page errors.
Anthropic-xAI Colossus-1: 220K GPUs, $1.25B/Month, and Rate Limits
Anthropic buys exclusive access to xAI's Colossus 1 cluster: 220K GPUs, $1.25B/month, and immediate Claude rate limit increases.
OpenAI + Dell: Codex On-Premises Architecture for Enterprise
OpenAI named Dell as its first non-hyperscaler Codex deployment path. Here's how the architecture actually works and who it targets.
Gartner Enterprise AI Coding Agents 2026: New Category, Four Leaders
Four Leaders, 12 vendors, one renamed category. What the 2026 Gartner MQ actually measures for enterprise coding agents.
Starlette BadHost: CVE-2026-48710 Auth Bypass in AI Agent Stacks
Starlette BadHost (CVE-2026-48710): a crafted Host header bypasses auth middleware. Unproxied AI agents at highest risk.
Grok Build CLI: Plan Mode, Skills, Connectors, and Pricing
xAI's Grok Build ships with Arena Mode, reusable Skills, and CLAUDE.md compat. Here's what developers need to know.
Codex CLI v0.134.0: History Search, MCP OAuth, and a Breaking Profile Change
v0.134.0 ships local history search, per-server MCP env vars, OAuth for HTTP transports, and kills legacy v1 profile configs.
Meta One AI Subscriptions: Tier Breakdown and Developer Implications 2026
Meta's first paid AI tiers arrive at $7.99 and $19.99/month. Here's what compute gating on Llama means for developers.
Robinhood Agentic Trading 2026: MCP, Sandbox Design, and Risk
Robinhood's MCP agentic trading beta: sandbox isolation, guardrails, and developer implications.
xAI Grok Build: Sub-Agents, MCP Compat, and the SWE-Bench Numbers
xAI shipped its terminal coding agent on May 14, 2026. Here's what the CLI actually does, where the benchmark numbers hold, and what $299/month buys.
Erdős Unit Distance Conjecture Disproved: Inside the OpenAI Proof
OpenAI's reasoning model disproved an 80-year-old geometry conjecture — verified by a nine-mathematician team including a Fields Medalist.
Microsoft Copilot Cowork: File Exfiltration via Prompt Injection
PromptArmor shows how a poisoned SKILL.md in OneDrive lets attackers silently pull M365 files — no approval dialog, no user alert.
Netflix AI Animation Stack: INKubator, InterPositive, and What's Next
Netflix quietly built two AI production units in March 2026. Here's how INKubator and InterPositive map together as an end-to-end pipeline.
AI Safety Law 2026: Illinois, California, and New York Compared
Illinois SB 315 goes further than CA and NY with mandatory third-party audits. Here's how the three laws differ and what it means for developers.
Google Managed Agents API: Sandbox, Skills, and Agentic Stack Analysis
One API call provisions a hosted Linux agent with persistent state and GCS mounts. Here's what developers need to know.
Mistral Custom Silicon: Inference Cost Math and the Feasibility Gap
Arthur Mensch hinted at chip design. Here's the inference economics behind the signal and why the startup feasibility gap is real.
vLLM v0.22.0 RC3: Multi-API-Server Timeout Fix Explained
RC3 patches a hard-coded 60s startup timeout in vLLM's multi-API-server subsystem — here's what changed and what operators must configure.
AI Marketing Claims and FTC Section 5: A Compliance Guide for 2026
The CMG Active Listening case sets the FTC's bar for AI capability and consent claims. What dev teams need to know.
The Real BadHost Risk: MCP Servers, vLLM, and the Proxy Gap
CVSS 6.5 misses the mark. Why MCP servers and proxy-less AI agent stacks face disproportionate exposure from BadHost.
Google AI Mode: U.S. Query Patterns, Agents, and Zero-Click Data 2026
I/O 2026 data shows 3× longer queries, 60% zero-click rate, and a new class of background agents. Here's the architecture.
Netflix INKubator: What Job Listings Reveal About the GenAI Stack
Netflix's AI animation studio emerged from job listings, not PR. Here's what the hiring data reveals about the pipeline architecture.
Google AI Mode 2026: What Agentic Search Means for Developers
AI Mode crossed 1B users at I/O 2026. Queries are 3× longer, background agents go live this summer. Here's what structurally changed.
Illinois SB 315 Explained: Who It Covers and What Devs Must Do by 2028
SB 315 passed 110-0. Who the $500M threshold covers, what five obligations apply, and when enforcement starts.
Gemini 3.5 Flash: Benchmarks, Pricing, and API Changes for 2026
Gemini 3.5 Flash is GA: 1M-token context, a breaking thinking_level change, and full pricing breakdown.
openai-codex b1 → b2: What the Same-Day Beta Fix Reveals
Two beta releases in under four hours. Here's what the b1→b2 patch cadence tells developers about SDK maturity and what to pin.
Google Beam, 다자간 통화로 확장: 3D 그룹 통화 실험의 작동 원리
Google I/O 2026 extended Beam to multi-person calls. Here's the AI pipeline, the $24,999 display, and where the gaps are.
과학을 위한 Gemini: Google I/O 2026이 연구자들을 위해 선보인 것들
Google's I/O 2026 AI research suite: literature triage, hypothesis tournaments, and ERA outperforming CDC forecasts.
중국의 AI 토큰 선물 계획: 2026년 상하이가 그리는 청사진
Shanghai Futures Exchange is prototyping AI token futures — forward contracts on LLM consumption costs. Here's the technical picture.
DiffusionBlocks: Sakana AI의 블록 단위 학습, ICLR 2026
DiffusionBlocks trains one residual block per step, reducing activation memory B× with competitive or better accuracy.
Robinhood MCP: AI 에이전트가 주식 거래와 결제를 직접 실행하는 방법
Robinhood opened its brokerage and card infrastructure to MCP-compatible AI agents. Here's what the implementation looks like technically.
langchain-perplexity 1.3.0: ChatPerplexity에 use_responses_api 추가
ChatPerplexity gains use_responses_api in 1.3.0: auto-routes to Perplexity's Agent API for real-time search.
Anthropic 실행 중 제약 블록: 캐시 안전 프롬프트 업데이트 2026
Mid-conversation constraint injection in v0.105.0 preserves prompt cache continuity across long inference runs.
Microsoft Copilot Cowork: 프롬프트 인젝션으로 M365 파일 유출
A 5-line poisoned Skills script silently exfiltrates SharePoint data via Copilot Cowork — no approval gate, no CVE, no patch.
Project Genie + Street View: 현실 세계 시뮬레이션, Genie 3에 탑재
Genie 3 generates interactive worlds from real Street View geometry. Waymo is already using it for rare-event training.
DeepMind의 러닝 가이드 에이전트: 시각장애 운동선수를 위한 온디바이스 Gemma 4
DeepMind's chest-mounted AI system lets blind runners navigate independently using dual-path on-device inference—no cloud, no tether.
Anthropic SDK 0.105.1 · 0.105.2: PyPI Trusted Publishing 긴급 패치
Two rapid patches followed Anthropic's 0.105.0 drop. Here's what broke, why, and which version to pin.
Claude Code 2026년 5월: 권한 버그 수정, /code-review --fix, MCP 인증
Seven builds in one week: four Bash/PowerShell sandbox bugs patched, /code-review --fix lands auto-apply, and a serious MCP auth credential leak is closed.
I/O 2026의 Gemini for Science: 각 연구 도구는 어떻게 작동하나
Three experimental AI research tools launched at I/O 2026. What Literature Insights, Co-Scientist, and AlphaEvolve each actually do.
Google Workspace I/O 2026: Docs Live, Gmail Live, Gemini Spark 총정리
Docs Live, Gmail Live, Gemini Spark, Sheets one-shot: I/O 2026 Workspace features and who gets access first.
Anthropic 0.105.0 심층 분석: 출력 귀속과 파일 용량 제한
v0.105.0 adds granular output-type attribution and configurable upload caps—here's what they do and when to use them.
vLLM v0.21.0 프로덕션 업데이트: KV 오프로드와 멀티 서버 포트 버그
v0.22.0 doesn't exist yet. v0.21.0 ships KV offload, spec decode, and a multi-server port bug still under review.
Claude Code v2.1.144–v2.1.154: 9일 만에 배포된 것들
Ten patches in nine days: pinned sessions, four security fixes, /code-review --fix, and skill-level tool gating.
SuperGrok 구독으로 Kilo Code에서 grok-build-0.1 사용 가능
SuperGrok and X Premium+ subscribers can now authenticate into Kilo Code and run grok-build-0.1 inside VS Code or JetBrains — no API key management required.
Codex CLI 0.134.0 & 0.135.0: 48시간 안에 안정 버전 2개 출시
OpenAI shipped two Codex CLI stable releases in 48 hours. What changed, what broke, and why the cadence matters.
Anthropic Python SDK 0.105: Opus 4.8 및 미드-세션 시스템 프롬프트
Three SDK releases in 7.5 hours ship claude-opus-4-8 support, mid-conversation system blocks, and finer output usage reporting.
xAI grok-build-0.1 API 공개 베타: 토큰 비용 및 SDK 지원
xAI's coding model exits the $299 CLI gate. Here's what the public API beta actually offers developers.
Grok Build, OpenCode·Kilo Code에 상륙: xAI의 13일 롤아웃
xAI shipped grok-build-0.1 to three developer tools in 13 days. Here's what each integration covers and how to pick the right surface.
Codex CLI 0.135.0: Doctor, Vim 텍스트 오브젝트, 그리고 0.136 알파
OpenAI's 0.135.0 stable is a diagnostics and polish cycle. What moved in the TUI, Vim mode, and remote transport.
Command A+ 2026: 벤치마크 결과, 인용 태그, 그리고 엔터프라이즈 적합성
Cohere's first open-weight frontier model: benchmark gaps, native citation design, and the enterprise sovereignty case.
grok-build-0.1: 모델 스펙, 가격 정책, 베타 출시 전말
xAI's grok-build-0.1 hit public beta in May 2026. Here's what the spec says — and what the caching incident revealed.
Claude Opus 4.8: 코딩 벤치마크와 에이전틱 업그레이드 2026
Anthropic ships Opus 4.8 with 69.2% SWE-Bench Pro, mid-conversation system messages, and adaptive thinking.
Codex CLI 0.135.0-alpha.2: 범위, 차이 분석, 그리고 릴리즈 노트 오류
Two alpha releases in three hours, 529 files changed. Here's what the diff says when the release notes page errors.
Anthropic-xAI Colossus-1: GPU 22만 개, 월 12.5억 달러, 그리고 요청 한도
Anthropic buys exclusive access to xAI's Colossus 1 cluster: 220K GPUs, $1.25B/month, and immediate Claude rate limit increases.
OpenAI + Dell: 기업용 Codex 온프레미스 아키텍처
OpenAI named Dell as its first non-hyperscaler Codex deployment path. Here's how the architecture actually works and who it targets.
Starlette BadHost: AI 에이전트 스택의 CVE-2026-48710 인증 우회 취약점
Starlette BadHost (CVE-2026-48710): a crafted Host header bypasses auth middleware. Unproxied AI agents at highest risk.
Netflix INKubator: 채용 공고로 드러난 GenAI 스택의 실체
Netflix's AI animation studio emerged from job listings, not PR. Here's what the hiring data reveals about the pipeline architecture.
구글 AI 모드 2026: 에이전틱 검색이 개발자에게 미치는 영향
AI Mode crossed 1B users at I/O 2026. Queries are 3× longer, background agents go live this summer. Here's what structurally changed.
일리노이 SB 315 해설: 적용 대상과 2028년까지 개발자가 해야 할 일
SB 315 passed 110-0. Who the $500M threshold covers, what five obligations apply, and when enforcement starts.
Gemini 3.5 Flash: 벤치마크, 가격, 2026년 API 변경 사항
Gemini 3.5 Flash is GA: 1M-token context, a breaking thinking_level change, and full pricing breakdown.
openai-codex b1 → b2: 당일 베타 픽스가 드러낸 것
Two beta releases in under four hours. Here's what the b1→b2 patch cadence tells developers about SDK maturity and what to pin.
AI developer tools and ecosystem news for developers and technical founders
