Top 3 on X today
- Jack Dorsey says his team built Dessa, a full-duplex speech model made so talking with AI sounds natural, with pauses, emotion and cadence.@jack · 34K likes
- Elon Musk is pitching Grok Bot as a life and business manager that can help with almost anything.@elonmusk · 25K likes
- Moonshot's new open-weight Kimi K3.1 Pro has a 1M-token context window and 104B active parameters, and targets full-project coding and 3D modeling.@Kimi_Moonshot · 23K likes
LiveChecked Oct 23, 10:45 PM ET
What's the latest in AI
New models, leaderboard moves, launches and funding, pulled straight from the labs and trusted feeds every two hours. Each headline links to its source.
- $237MAI drug development startup Pathos AI closes $237M investmentsiliconangle.com
- FundingThe Week’s 10 Biggest Funding Rounds: AI Drives A Varied Lineup Of Large Financingsnews.crunchbase.com
- OAIRapidly scaling online storage to serve over 1 billion ChatGPT users (Part II)openai.com
- LitPivot: Developing research ideas through dynamic contextualization within the literature landscapehuggingface.co
- OAISecuring our frontier model training environmentopenai.com
- $30BReport: OpenAI closing on $30B ‘bridge’ funding round to tide it over before IPOsiliconangle.com
- OAIOpenAI and Poland’s NASK partner to advance AI safety and securityopenai.com
- Does Your Model Run on AWS Trainium? Look It Uphuggingface.co
- Turning Database Grading into RL Rewards: Post-Training Agents with ThinkingBoxhuggingface.co
- $30BOpenAI working on new $30B raise at $1.25T valuation ahead of IPOtechcrunch.com
- OAIBuild more responsive apps with the Decisions APIopenai.com
- GOOGBeyond tree cover: mapping the world’s forest typesdeepmind.google
Top models: LMArena TextFull board
| # | Model | Score |
|---|---|---|
| 1 | Gemini 4 Argon (High) google | 1526 |
| 2 | Claude Opus 5.5 (High) anthropic | 1516 |
| 3 | Claude Fable 5.1 (Max) anthropic | 1513 |
| 4 | Claude Opus 5 (Max) anthropic | 1508 |
| 5 | Claude Opus 5 (High) anthropic | 1503 |
| 6 | Claude Opus 4.6 (High) anthropic | 1501 |
| 7 | Muse Wonder (Max) meta | 1501 |
| 8 | Gemini 3.8 Flash (High) google | 1498 |
| 9 | Claude Opus 4.6 anthropic | 1497 |
| 10 | Muse Spark 1.3 (Max) meta | 1491 |
Board published Oct 22, 2026. Source: LMArena, CC BY 4.0.
Today's digestRead it
Oct 23, 2026: 1 new model listing, 5 lab posts, 2 funding stories, 9 launches and releases.
Latest model releasesAll models
MoonshotAINew Kimi K3.1 Pro 1.0M context$2.00 / $10 MiniMaxNew MiniMax M3.1 Flash 1.0M context$0.15 / $0.60 DeepSeekNew DeepSeek V4.1 Flash 1017 1.0M context$0.15 / $0.60 ElorianNew Glimpse 262K context$0.60 / $3.00 Reflection Beam 1.0 501B A23B 1.0M context$0.25 / $1.00 Inception Mercury 3 1M context$0.100 / $0.35 Moondream Moondream 3 Preview 32K context$0.30 / $2.50 Moondream Moondream 3.1 32K context$0.30 / $2.50
Date first listed on OpenRouter. Price is US dollars per 1M input / output tokens.
FundingMore funding
- $237MAI drug development startup Pathos AI closes $237M investmentsiliconangle.com, Oct 23
- Oct 23The Week’s 10 Biggest Funding Rounds: AI Drives A Varied Lineup Of Large Financingsnews.crunchbase.com, Oct 23
- $30BReport: OpenAI closing on $30B ‘bridge’ funding round to tide it over before IPOsiliconangle.com, Oct 22
- $30BOpenAI working on new $30B raise at $1.25T valuation ahead of IPOtechcrunch.com, Oct 22
- $33MNirva raises $33M for a pendant that only listens to youtechcrunch.com, Oct 22
- $15BDeepSeek reportedly raises $15B days after US restrictionstechcrunch.com, Oct 22
LaunchesMore launches
- 2342DeepInfer: High-performance inference framework for DeepSeek models891 comments on Hacker News
- 748Kimi K3.1 Pro and K3.1 Pro Highspeed467 comments on Hacker News
- 619Every AI model has the same "controversial" take on humans594 comments on Hacker News
- 597Ask HN: If the models are so good, why doesn't OpenAI just start 100 companies?667 comments on Hacker News
- 561Claude Duet531 comments on Hacker News
- 489OpenShell, A safe, private runtime for fleets of autonomous AI agents213 comments on Hacker News
ResearchMore papers
- 223WeVisOne-VL: Unified Visual Perception from Objects to PixelsarXiv 2610.31329, Oct 21
- 165Making MLLM Latents Matter: Factor-Aware Predictive Learning for Cinematic Video GenerationarXiv 2610.30926, Oct 22
- 152Deep Search in Real-World Web EnvironmentsarXiv 2610.30181, Oct 18
- 74Editing Images into Reusable Layers via Compositional Edit-Prior DistillationarXiv 2610.24738, Oct 16
- 55IQuest-Q1: Advancing Foundation Capabilities for Agentic CLI SystemsarXiv 2610.30949, Oct 21
- 47Toward Recursive Self-Improvement in Physical AIarXiv 2610.29677, Oct 20
Upvotes on Hugging Face Papers.
Latest versions of AI toolsAll answers
| Tool | Latest release | Date |
|---|---|---|
| Claude Code Anthropic | v2.1.310 | Oct 23 |
| Ollama Ollama | v0.40.10 | Oct 23 |
| Codex CLI OpenAI | rust-v0.164.0 | Oct 23 |
| LangChain LangChain | langchain-core==1.7.0 | Oct 22 |
| Gemini CLI Google | v0.65.0 | Oct 20 |
| Transformers Hugging Face | v5.20.2 | Oct 19 |
| vLLM vLLM project | v0.32.0 | Oct 18 |
| OpenAI Agents SDK (Python) OpenAI | v0.24.0 | Oct 14 |
From each project's GitHub releases.
Trending open models
| # | Model | Likes |
|---|---|---|
| 1 | deepseek-ai/DeepSeek-V4.1-Flash-1017 image-text-to-text | 2,594 |
| 2 | RecursiveSI/flint text-generation | 690 |
| 3 | black-forest-labs/flux-3-image-klein-comfy image-to-image | 1,411 |
| 4 | OpenMOSS-Team/MOSS-Image-SFT text-to-image | 656 |
| 5 | Lightricks/LTX-2.5 image-to-video | 9,039 |
| 6 | XHToken/Spark-X2.6-4B text-generation | 417 |
| 7 | fishaudio/drama-3 text-to-speech | 948 |
| 8 | OpenMOSS-Team/MOSS-Image-Base text-to-image | 521 |
Trending on Hugging Face.
What is the latest model from each lab?All answers
OpenAIOAI GPT-6.1 Sol Pro Top on LMArena Text#30 AnthropicANT Claude Haiku 5.5 Top on LMArena Text#2 GoogleGOOG Gemini 4 Argon Top on LMArena Text#1 xAIXAI Grok 4.7 Top on LMArena Text#51 MetaMETA Muse Spark Open Top on LMArena Text#7 Mistral AIMIS Mistral Large 4 Top on LMArena Text#92 DeepSeekDSK DeepSeek V4.1 Flash 1017 Top on LMArena Text#37 Qwen (Alibaba)QWEN Qwen3.8 Max Prime Top on LMArena Text#15 Moonshot AIKIMI Kimi K3.1 Pro Top on LMArena Text#25 NVIDIANVDA Nemotron 3.5 Super VL Top on LMArena Text#63