AGI Pulse
AGI STATUS

AGI is coming

See how major breakthroughs have pushed the frontier of intelligence since 1950.

Tracking live
Long-term AGI progress curve; open the full curve for milestone details
Latest signal 2026-10-31 20 key changes this week Tracking 1950—now
Open full curve →

What changed this week

Key changes this week · 20
10-30
OpenAI launches Private Intelligence: encrypted safety-review records stay in customer-owned cloud storage with no human access, and confidential-computing Private Inference enters API preview
ProductOpenAIOpenAI
10-29
Mistral opens weights for Large 4 (Le Chonk): 1T multimodal open-weight model with coding/cyber gains; vLLM Day-0
Model releaseMistral AIagipulse
10-28
Google DeepMind announces Gemini 4 Carbon: AA Intelligence Index v5 #1; SOTA cyber find/exploit/patch benches; Fairwind limited rollout after Argon
Model releaseGoogle DeepMindagipulse
10-28
NVIDIA opens Nemotron 3.5 Transcribe: promptable open-weight multilingual ASR with diarization, translation, and audio Q&A; Granary v2 data pipeline
Model releaseNVIDIAagipulse
10-27
Google DeepMind announces Genie 4: real-time interactive world model with native audio, Gemini Omni steering; research preview then Ultra/Project Genie + API early access
Model releaseGoogle DeepMindagipulse
10-26
Cloudflare discloses cross-request KV cache leak on Workers AI GLM-5.3 (Aug 27–Sep 9); 155 long-context replies affected; SGLang v0.5.22+ fixes
NewsCloudflareagipulse
10-26
OpenAI launches GPT-6.1 Terra: mid-tier visual/computer-use model; API $1/$0.05/$5; ChatGPT Work+Codex; Critical cyber / High bio-chem
Model releaseOpenAIagipulse
10-26
Anthropic, PBC files public S-1 for Nasdaq “ANT” IPO: H1’26 revenue $16.3B, PBC+LTBT controlled structure; share count and price TBD
IndustryAnthropicagipulse
10-15
Meta releases Muse Spark Open weights: 816B-A42B multimodal model with 1M context, built for coding and agents
Model releaseMetaagipulse
10-15
Black Forest Labs opens FLUX 3 Image Klein: 5.6B open weights for text-to-image, editing and multi-reference (native 4K)
Model releaseBlack Forest Labsagipulse
10-15
NVIDIA opens Nemotron 3.5 Super VL (120B-A12B multimodal agents) and Bolt 74B-A9B 24GB for local agents; Bolt E4B due Oct 23
Model releaseNVIDIAagipulse
10-15
StepFun opens Step 5 Preview weights: 600B sparse MoE (27B active), 1M context, FP8/BF16 on Hugging Face and ModelScope
Model releaseStepFunagipulse
10-13
Google opens Gemini 4 Argon globally: paid Gemini API, Antigravity for Ultra, Gemini app and Search AI Mode for Pro/Ultra; intro price $2/$10 per million tokens
Model releaseGoogleagipulse
10-12
Meta launches Muse Wonder and Muse Wonder Full: frontier agent models with Active Memory and Active Focus; API and Muse Code today, Muse assistant for subscribers next
Model releaseMetaagipulse
10-08
Google Cloud launches the Gemini agent for work: one agent for questions, documents and code that runs for days, spawns sub-agents and routes jobs between Gemini and Claude
ProductGoogleagipulse
10-08
CrowdStrike: attacker used open-source AI pentest agent ARTEX with DeepSeek and Claude Code to breach South Korean banks
NewsCrowdStrikeCrowdStrike Intelligence
10-07
Epoch AI's InnovationEval: frontier agents fall far short of a human post-training advance, and cherry-pick their results
PaperEpoch AIEpoch AI
10-07
Mistral Large 4 preview: a 1T-parameter open-weight model pitched on refusal-free cyber work, weights due end of month
Model releaseMistralMistral
10-07
GPT-6 comes to everyday ChatGPT with Intelligent UI; Free users follow a day later on Luna
ProductOpenAIOpenAI
10-06
OpenAI releases math results from an internal frontier model: 722 manuscripts from about 4,000 open problems, including a Hodge Conjecture case and a zeta zero-free region, with Lean proofs for about 162
PaperOpenAIOpenAI
More →
Latest · 18 Site updated 10-31 12:46 (UTC+8)
10-31
Anthropic: Claude automates search for subtle emergent-misalignment training data that fools production-like monitors
PaperAnthropicAnthropic
10-30
OpenAI rallies amicus support for fair use in NYT copyright case as summary-judgment amicus deadline closes: US government, libraries, blind advocates, Authors Alliance back AI training
NewsOpenAIOpenAI
10-30
Google's Project Suncatcher satellite returns first AI output from a data-center-class TPU in orbit: onboard Gemma 4 writes a haiku; twin interconnected satellites next in 2027
NewsGoogleGoogle
10-30
OpenAI commits to a semiannual Community Safety Report and tightens ChatGPT violence and teen weapons limits; 200+ law-enforcement referrals in Q3
NewsOpenAIOpenAI
10-30
Anthropic renames Claude Max 20x to Ultra ($200/mo) and removes the 5-hour usage limit for verified Ultra accounts
ProductAnthropicAnthropic
10-30
OpenAI, Google and Microsoft back a shared IETF robots.txt token for AI-training opt-outs; Bing also splits training from grounding, effective 2027
IndustryOpenAIOpenAI
10-30
OpenAI releases GPT-6 Rosalind for trusted-access life-sciences research, priced at $15/$75 per million tokens, 3x the prior Rosalind models
Model releaseOpenAIOpenAI
10-30
humans& releases &Tutor: RL-trained on GLM-5.3 against its Persimmon learner model, it beats Claude Sonnet 5.5 and GPT-6 Astra on TutorMoments by default
Model releasehumans&humans&
10-30
UK CMA opens competition probe into Meta terms that could let it scrape publishers’ sites for AI
NewsUK CMACompetition and Markets Authority
10-30
Microsoft open-sources Mage-VL-Cowork: 4B hybrid GUI+terminal computer-use agent hits 66.0% on OSWorld-Verified
Model releaseMicrosoftMicrosoft
10-30
Microsoft AI ships MAI Image 2.7 (AA image-edit #2, Arena Edit #3) and MAI Cyber 1.1 Flash (−18% MDASH scan cost)
Model releaseMicrosoft AIMicrosoft AI
10-30
CrowdStrike and Microsoft: state hackers go agentic — China's VAPOR PANDA hit 200+ orgs in 11 days with ARTEX + DeepSeek; North Korea has operationalized AI
NewsCrowdStrikeCrowdStrike
10-30
Cohere opens CAD 7B Series E at CAD 35.7B valuation, led by Schwarz Group, with Canada taking an up-to-CAD-1.6B equity stake
IndustryCohereCohere
10-30
OpenAI expands multilayer monitoring to detect frontier model misalignment across training, evals, and deployment
NewsOpenAIOpenAI
10-29
OpenAI updates Model Spec around five values: intent faithfulness, harm minimization, honesty, agency, care
NewsOpenAIOpenAI
10-29
FuriosaAI closes $620M Series D at $2.9B valuation; RNGD inference cards in mass production (20k cards in 2026)
IndustryFuriosaAIagipulse
10-26
OpenAI Decisions API goes GA with gpt-6-luna: bounded answers, up to 10x faster than Responses
ProductOpenAIagipulse
10-26
Claude Enterprise unifies Chat and Cowork: cloud sessions go GA, egress off by default, with Efficient mode
ProductAnthropicagipulse
More →
See full timeline →