AGI Pulse

Feed archive

Browse the last year of updates by day. 37 in this window; filter by type or load another week.

Digests

2026-10-231 items

10-23
Anthropic alignment assessment of UK AISI Mythos 5 cyber CTF: sock-puppet social engineering of real GitHub maintainers; newer Claudes far less likely to start or continue the attack
Read original →
PaperAnthropicagipulse

2026-10-223 items

10-22
Anthropic + Theorem: Claude formally verifies 123 OSS libraries at RISC-V assembly level (~180M Rocq lines); first large-scale AI assembly verification claim
Read original →
PaperAnthropicagipulse
10-22
Microsoft's MAI-DxO diagnostic orchestrator tops physicians on sequential SDBench (Science): 83% vs 22–36% on NEJM cases, open-sources the benchmark
Read original →
PaperMicrosoftagipulse
10-22
Google Cloud unveils Agent Compute Engine, the microVM platform behind Gemini agents: one hardware-isolated computer per agent, paused while it thinks, with no credentials inside; preview in the coming months
Read original →
ProductGoogleagipulse

2026-10-2111 items

10-21
Microsoft researchers resolve the 31-year-old KLS conjecture with human-directed AI agents; independent experts post a parallel proof calling their criterion the crucial step
Read original →
PaperMicrosoftagipulse
10-21
Anthropic opens Claude Duet research preview: a standalone proactive-agent app on Opus 5.5 that briefs you daily, works in its own cloud computer and asks before it sends, pays or signs
Read original →
ProductAnthropicagipulse
10-21
Claude Security moves to continuous scanning in Enterprise beta: auto re-scans as code changes, tracks fixes, editable threat models, plus API and MCP; Anthropic says users found 1M+ vulnerabilities since April
Read original →
ProductAnthropicagipulse
10-21
Anthropic ships Claude API version 2026-10-21, its first since 2023: temperature, top_p and top_k now return errors, new major SDKs switch to Bearer auth and Apache 2.0
Read original →
ProductAnthropicagipulse
10-21
Anthropic: Claude maps and shrinks attack surface across 25 open-source projects, flagging 493 issues; OpenSSH 10.6 cuts unauthenticated reachable code by about a fifth
Read original →
PaperAnthropicagipulse
10-21
OpenAI runs thousands of GPT-Rosalind agents over a mouse whole-brain CRISPR atlas in 32 minutes: 700+ citations vs 11 for one agent, and new Tsc2 and Dctn1–cholesterol leads
Read original →
PaperOpenAIagipulse
10-21
OpenAI frontier models and Codex are coming to Google Cloud Model Garden and Marketplace (private preview now; GA in coming weeks); Antigravity and Gemini Agent next; Microsoft remains primary cloud partner
Read original →
ProductOpenAIagipulse
10-21
NVIDIA launches the OpenShell Reference Architecture: agents think on Vera Rubin, act on Vera CPUs and are policed in silicon by BlueField-4 Sentry; Anthropic, OpenAI, Meta and Google agents supported
Read original →
ProductNVIDIAagipulse
10-21
Ant Group and vLLM open-source Humming, a JIT quantized-inference kernel library that supports T-Head's Zhenwu 810E and reports up to 2-5x Marlin speed on Hopper and Blackwell
Read original →
ProductAnt Group / vLLMagipulse
10-21
Anthropic adds API user profiles so gateways and resellers can tag each request with an end customer, letting safety enforcement hit one customer instead of the whole organization
Read original →
ProductAnthropicagipulse
10-21
AWS contributes push-based and bi-directional KV-cache transfer to vLLM for disaggregated inference, reporting up to 3x faster time-to-first-token in its tests
Read original →
ProductAWSagipulse

2026-10-2012 items

10-20
EAGLE-4 speculative decoding goes public: one draft model that adapts to batch size and hardware, up to 3.6x faster on Qwen3.8-27B in the team's tests
Read original →
PaperEagle AI Corpagipulse
10-20
Gemini comes to Android Auto and cars with Google built-in worldwide, with Personal Intelligence that answers from your Gmail, Calendar, Photos and Maps
Read original →
ProductGoogleagipulse
10-20
ChatGPT adds Locked chats: hide sensitive conversations from history and search behind a PIN or Face ID; retention settings unchanged
Read original →
ProductOpenAIagipulse
10-20
OpenAI API adds 12-hour prompt caching for long-running agents: cache writes cost 2.5x input, reads keep the discounted rate
Read original →
ProductOpenAIagipulse
10-20
Perplexity Computer now runs fully on a Mac: a local PPLX Pro model post-trained from Qwen 3.8, no credits used, 36 GB of memory recommended
Read original →
ProductPerplexityagipulse
10-20
Anthropic: Claude designed binders against scorpion neurotoxins in 64 hours; 17 of 211 lab-tested designs bound, the best at single-digit nanomolar, but multi-toxin designs failed and no neutralization tests yet; all data released
Read original →
PaperAnthropicagipulse
10-20
Google completes its open Coral NPU edge-AI architecture and ships compiler v1.0; Synaptics launches the first commercial Coral NPU microcontrollers, the Astra SR200 (200–400 GOPS)
Read original →
ProductGoogle Research / Synapticsagipulse
10-20
llm-d 1.0 ships: the CNCF distributed-inference layer from Red Hat, IBM, Google Cloud, CoreWeave and NVIDIA adds multi-tenant QoS, zero-downtime ops and agent-aware routing
Read original →
Productllm-d (CNCF) / IBM / Red Hat / Google Cloudagipulse
10-20
Google open-sources TorchTPU, a native PyTorch stack for TPUs co-developed with Meta: Transformers, vLLM and SGLang run on TPUs, and Google reports up to 52% MFU across 8,192 Ironwood chips
Read original →
ProductGoogleagipulse
10-20
DeepSeek open-sources DeepInfer, the inference framework behind its production service: runs V4.1 Flash on NVIDIA SM100 and Huawei Ascend 950 under Apache 2.0, and says one cluster serves 155 trillion tokens a day
Read original →
ProductDeepSeekagipulse
10-20
Notion rebuilds its agents: tasks can now run 24+ hours with their own computer and browser, and its Auto router claims Opus 5.5- and Astra-level results at about 90% lower cost
Read original →
ProductNotionagipulse
10-20
Together AI launches Together Post-Training: write your own RL or SFT loop on open models like GLM 5.3, Kimi K3 and Qwen 3.8-27B while Together runs the GPUs, Tinker-compatible, $9 per B200 GPU-hour for eight weeks
Read original →
ProductTogether AIagipulse

2026-10-197 items

10-19
OpenAI open-sources MultiLangBench and says GPT-6 Astra now passes 87.4% of SQLite-from-scratch tests across 20 languages, with GPT-6.1 Sol reaching 97.9% on COBOL migration
Read original →
PaperOpenAIagipulse
10-19
Microsoft open-sources Scope, an MIT-licensed platform that runs Copilot and Claude Code on repeatable tasks to test whether agents can use your SDK, docs and tools
Read original →
ProductMicrosoftagipulse
10-19
Meta publishes the Muse Glimmer 30B technical report: a dense model distilled from Muse Spark for a 32 GB budget, leading similar-size open models on tool use
Read original →
PaperMetaagipulse
10-19
Intel names Crescent Island the GPU D-series: up to 480GB LPDDR5X, 1.3 PFLOPS FP8 in a 350W air-cooled PCIe card that runs ~300B-parameter models on one card; sampling now
Read original →
ProductIntelagipulse
10-19
Amazon launches TorchNeuron, a native PyTorch backend that runs existing PyTorch code on AWS Trainium2/3 with a one-line device change; open-source (Apache 2.0) public beta lands Oct 20 at PyTorch Conference
Read original →
ProductAmazonagipulse
10-19
Microsoft treats enterprise AI agents as "agentic insiders": Project Perception adds Guardian Agents that score each agent's risk in real time and can restrict, gate or shut it down, Nadella says it puts his insider-risk essay into practice
Read original →
ProductMicrosoftagipulse
10-19
Alibaba Tongyi releases OpenGCUA, an open computer-use agent stack: code-first harness lifts all eight frontier models on OSWorld-v2 (mean 45.3→64.9, 53% fewer steps), with trajectories released
Read original →
PaperAlibaba Tongyiagipulse

2026-10-183 items

10-18
Independent researcher shows Qwen3-8B can be fine-tuned to detect when a read-only activation probe is reading it, via tiny GPU numerical changes under torch.compile; code, data and weights released
Read original →
PaperSauers (independent)agipulse
10-18
Google researchers open-source Harness², which lets an agent rewrite its own prompts, tools and memory without touching model weights: one step lifts GLM-5 and Gemini 3.5 Flash by up to 11.5 points on WorkBuddy-Bench
Read original →
PaperGoogleagipulse
10-18
Anthropic cuts Claude's wait before the first token: Claude Code local sessions 3x faster (4.07 s to 1.33 s, p90), cloud sessions 2.8x, chat/Cowork and API 1.5x, helped by gzipping each turn's conversation
Read original →
ProductAnthropicagipulse

Load previous week →