AGI Pulse

Feed archive

Browse the last year of updates by day. 51 in this window; filter by type or load another week.

Digests

2026-10-234 items

10-23
Moonshot launches Kimi K3.1 Pro: 2.8T MoE open-weight flagship (1M ctx); API today, full weights by Nov 4; AA Index 55 (#4)
Read original →
Model releaseMoonshotagipulse
10-23
NVIDIA Nemotron 3.5 Bolt E4B open on Hugging Face (Gemma-4-E4B post-train, ~4GB GGUF); Build and Ollama day-0
Read original →
Model releaseNVIDIAagipulse
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
10-23
MiniMax M3.1 Flash open weights on Hugging Face (~428B/23B MoE, 1M ctx); vLLM and Together day-0
Read original →
Model releaseMiniMaxagipulse

2026-10-227 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
Canva launches Craft, its first in-house image model: DesignArena #4 debut, two-stage plan-then-render, live in Dream Lab and Public API
Read original →
Model releaseCanvaagipulse
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
Hugging Face open-sources Carbon-X, a 14B Apache-2.0 model you can chat with about DNA: averages 66.7% on five DNA benchmarks vs GPT-6.1 Sol's 67.5% at ~1/1,800 the cost, runs on one 80 GB GPU
Read original →
Model releaseHugging Faceagipulse
10-22
miHoYo and Tsinghua open GameScaling: eight Apache-2.0 game-playing models (0.8B–27B) that output keyboard-and-mouse actions; 27B tops Minecraft specialists and 9B beats frontier models in real-time play
Read original →
Model releasemiHoYoagipulse
10-22
iFlytek launches Spark X2.6: a 293B-A30B MoE with 1M context that it says was trained entirely on Chinese chips, priced at ¥0.9/¥3.6 per million tokens; only the 4B sibling has open weights
Read original →
Model releaseiFlytekagipulse
10-22
Baseten's Base Labs open-sources decision models mammoth-1 (27B) and elephant-1 (9B) under MIT: mammoth-1 tops the Jev Decision Index at 71.24, elephant-1 ties Liquid's closed d1-ultra on Artificial Analysis
Read original →
Model releaseBasetenagipulse

2026-10-217 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
Cohere launches Rerank 5 Pro and Fast: tops BRIGHT at 82.2 (+5.8 over Voyage Rerank 3), adds Context Focus passage pruning that cut agent tokens 35% while raising accuracy, and probability-calibrated relevance scores
Read original →
Model releaseCohereagipulse
10-21
Recursive releases Flint, an Apache 2.0 open-weight 40B-A2.5B MoE for local coding agents: self-reported 74.2% SWE-bench Verified and 150+ tokens/s on a 64GB RTX Spark, built with its Eureka Machine automated-research system
Read original →
Model releaseRecursiveagipulse
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
NVIDIA opens Kumo foundation models for tabular business data and RFD4-Proteina, a protein design model with the Baker Lab it says is 13x faster than RFdiffusion3
Read original →
Model releaseNVIDIAagipulse
10-21
PUBG maker KRAFTON open-weights Raon-1 mini preview, its first LLM trained from scratch: a 30B-A4B MoE built on Kimi- and DeepSeek-derived attention, Apache 2.0
Read original →
Model releaseKRAFTON AIagipulse

2026-10-2015 items

10-20
humans& releases Ampersand v0.1, a 'collaborator' model RL-trained from GLM 5.3 to help groups reach shared goals, via an Agents API
Read original →
Model releasehumans&agipulse
10-20
ElevenLabs launches Eleven Realtime: v1 Mini and v1 Max speech-to-speech models for voice agents, with an OpenAI-compatible Realtime API
Read original →
Model releaseElevenLabsagipulse
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
UCSD's FastVideo open-sources FastH3 Omni, a distilled MiniMax-H3 that makes reference-guided video with sound up to 12x faster and runs on an RTX 5090
Read original →
Model releaseHao AI Lab (UC San Diego)agipulse
10-20
NVIDIA releases Cosmos3-H-Dreams, an open real-time surgical world model trained on clinical gallbladder surgery that runs at 30 fps on one GPU
Read original →
Model releaseNVIDIAagipulse
10-20
NVIDIA GEAR releases Flora, open models that caption and score robot and human videos to train GR00T; weights are non-commercial
Read original →
Model releaseNVIDIAagipulse
10-20
Radical Numerics launches Omnii-1 genome model and says its AI-designed phages were given to a patient with a drug-resistant infection; outcome not disclosed
Read original →
Model releaseRadical Numericsagipulse
10-20
Gradium open-sources Phonon, on-device text-to-speech models under 50M parameters that run with no GPU; only the English 31m model is licensed for commercial use
Read original →
Model releaseGradiumagipulse
10-20
Bria releases Fibo 2.0, an 8B image model that Artificial Analysis ranks #1 among open-weights image editors; weights are non-commercial only
Read original →
Model releaseBria AIagipulse
10-20
Mistral launches Voxtral 3: 50-language transcription, expressive speech in 35+ languages and a LiveKit-based voice agent SDK, API and self-hosted only
Read original →
Model releaseMistral AIagipulse
10-20
MemTensor open-sources MemT-1.0, memory-native models built on DeepSeek-V4-Flash and Qwen3.6 that keep a fixed-size memory inside the model instead of resending history; text-memory systems still lead overall
Read original →
Model releaseMemTensoragipulse
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
Microsoft Research releases a public preview of BioEmu2, an MIT-licensed open-weights foundation model for biomolecular structure and dynamics covering proteins, nucleic acids and ligands
Read original →
Model releaseMicrosoft Researchagipulse
10-20
Skyfall AI unveils SKY-001, an "engineering world model" for CAD and BIM design: it claims 51-125x the speed of GPT-6 Astra and other frontier models at 150-210x lower cost, early access by waitlist only
Read original →
Model releaseSkyfall AIagipulse
10-20
Ant Group's Ant Ling open-sources Ming-3.0, five MIT-licensed speech models: a 0.5B TTS that tops Artificial Analysis' open-weights voice-cloning arena, a 126B-A6.5B audio-understanding MoE and a dialect-aware ASR
Read original →
Model releaseAnt Group (inclusionAI)agipulse

2026-10-1915 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
Viggle releases JST-2, a physics-native world model that builds 3D scenes in about a second and rig-free animated characters, free in its PINOC app and via API
Read original →
Model releaseViggleagipulse
10-19
Amnon Shashua's doubleAI launches Maxwell, a 'code execution model' that rewrites running CPU and GPU code for speed and formally proves each change
Read original →
Model releasedoubleAIagipulse
10-19
Ex-NVIDIA GR00T team's Dream Labs launches Dream-1 robot model with a public API from $3/hour and a $90M seed round
Read original →
Model releaseDream Labsagipulse
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
Black Forest Labs and fal launch FLUX 3 Max video model: 10-second clips in about 11-12 seconds, joint #1 in Artificial Analysis' preliminary image-to-video ranking
Read original →
Model releaseBlack Forest Labs / falagipulse
10-19
Cloudflare previews Clairic, a System One extraction model on Workers AI that returns typed values, estimates and forecasts in one pass without generating tokens; open weights planned
Read original →
Model releaseCloudflareagipulse
10-19
Inception releases Mercury 3, a diffusion LLM it says beats GPT-6 Luna and Haiku 5.5 at 1,000+ tokens/s with 1M context; $0.20/$0.70 per 1M tokens, half off at launch
Read original →
Model releaseInceptionagipulse
10-19
MiniMax officially launches M3.1 Flash, reveals Space Bunny as its early version, publishes $0.15/$0.60 pay-go pricing and agentic benchmarks; open weights promised later
Read original →
Model releaseMiniMaxagipulse
10-19
GoTo releases a 27B Indonesian-language model with open weights on Hugging Face: Gemma 3 continued pre-training with AI Singapore, six languages incl. Javanese, Sundanese, Batak and Balinese, 128K context
Read original →
Model releaseGoToagipulse
10-19
ACE Robotics releases ACE-Brain-0.6, a 9B embodied model fine-tuned from Qwen3.5-9B on ~1M grounding samples, claiming to beat GPT-6 Astra on six of eight spatial-grounding benchmarks
Read original →
Model releaseACE Roboticsagipulse
10-19
MiniMax makes M3.1 Flash generally available in its API (1M context, multimodal, five thinking-effort levels) and starts retiring M2.5/M2.7, routing plan users to the new model
Read original →
Model releaseMiniMaxagipulse
10-19
VAST open-sources EdenLive: real-time generative rendering where a game engine keeps world state and a 5B/14B video model paints the frames, 24+ FPS at 720P on one RTX 5090 (self-reported)
Read original →
Model releaseVASTagipulse
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
10-19
Tencent Hunyuan open-sources Hy Memory 2.0: a ~3B-active memory-writer model (Hy-Memory-30B-A3B) that edges Claude Opus 5 in its own memory benchmarks, plus an SDK and MemHub cross-tool memory app
Read original →
Model releaseTencent Hunyuanagipulse

2026-10-182 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

2026-10-171 items

10-17
NVIDIA releases Nemotron Labs Dory1.0-4B on Hugging Face, a looped 4B model whose single checkpoint trades inference loops for accuracy (SWE-bench Verified 50.3→61.8 at 1→4 loops, self-reported; 256K context, OpenMDW-1.1)
Read original →
Model releaseNVIDIAagipulse

Load previous week →