AI/TLDR — every new AI model, tool, repo & paper
The latest AI releases, refreshed every 2 hours and explained in plain English.
What AI shipped today?
In the last 24 hours AI/TLDR tracked 19 new AI releases, including Xing4.0-29B-A4B — China Telecom's agent model trained on Ascend NPUs, Claude Code 2.1.275 — claude.ai skills and plugins sync to the terminal and Codex CLI 0.155.0 — voice conversations arrive in the terminal. AI/TLDR is an AI release tracker that follows new AI models, open-source tools, papers, datasets and benchmarks — refreshed every 2 hours from verified primary sources and explained in plain English.
AI Release Index — live stats on AI releases · Learn AI
- Xing4.0-29B-A4B — China Telecom's agent model trained on Ascend NPUs
Xing4.0-29B-A4B is an Apache-2.0 mixture-of-experts model with 29B parameters and 4B active per token. It scores 75.00 on SWE-bench Verified and 57.50 on Terminal-Bench 2.1, and was trained entirely on Ascend 910C chips.
- Claude Code 2.1.275 — claude.ai skills and plugins sync to the terminal
Claude Code 2.1.275 syncs the skills and plugins enabled on your claude.ai account into terminal sessions signed in with it, and adds a key that sends every queued message at once. A same-day 2.1.276 fixes a gateway regression.
- Codex CLI 0.155.0 — voice conversations arrive in the terminal
Codex CLI 0.155.0 adds experimental /voice conversations with live transcripts and microphone controls, Touch ID checks on MCP requests, and task hiding, archiving and deletion in the agents overview.
- Qwen3.8-Omni-Flash — Alibaba's omni model gets a 1M-token context
Qwen3.8-Omni-Flash is Alibaba's omnimodal model that reads text, images, audio and video in a 1M-token context. Alibaba says audio input costs over 98% less per hour than Qwen3.5-Omni-Plus, and average scores rose 25% across 29 tests.
- OpenAI publishes misalignment reports — six cases from its own training runs
OpenAI's Model Misalignment Reporting Framework sets out how the company flags, reviews and publishes cases where a model departs from its instructions. Six reports landed with it, all from reinforcement learning runs on unreleased models.
- Two Minute Papers — 'DeepSeek Just Made AI Memory 4x Smaller!'
Two Minute Papers posted 'DeepSeek Just Made AI Memory 4x Smaller!' on 18 September 2026. The subject is DeepSeek-V4.1-Flash's KV cache work, which cuts the global cache to 890 bytes per token — about a quarter of DeepSeek-V4-Flash.
- Wes Roth — 'OpenAI's Astra class model JAILBROKE ITSELF...'
Wes Roth posted 'OpenAI's Astra class model JAILBROKE ITSELF...' on 18 September 2026. The subject named in the title is OpenAI's new misalignment reports, one of which found 27 training summaries carrying jailbreak-style instructions.
- Bonsai 2 27B — ternary model keeps 98.2% of full precision in 5.9 GB
Bonsai 2 27B is an Apache-2.0 ternary model built from Qwen3.8 27B. It stores every weight as -1, 0 or +1 at 1.76 effective bits, scores 83.9 overall, and keeps 98.2% of the full-precision model with a 262K-token context.
- SoL-Pi — NVIDIA's harness extension cuts coding-agent tokens by about half
SoL-Pi is an MIT-licensed extension for the Pi coding agent from NVIDIA's research lab. It packages four token-saving mechanisms found by an automated research loop, cutting token use 45-49% while keeping about 94% of Pi's EdgeBench score.
- Anthropic publishes pace metrics — Claude leads 26% of its own AI research
Anthropic proposes three measurements frontier labs could publish about their own AI development and reports its August 2026 numbers: Claude leads 26% of Anthropic's AI R&D work, and about 6% of AI R&D compute goes to safety.
- Astra for Law — OpenAI ties GPT-6 Astra to a 230M-URL legal index
Astra for Law is OpenAI's legal setup for GPT-6 Astra: a search index of more than 230 million URLs of US law plus 26 partner plugins. OpenAI reports 54% accuracy on 200 legal research questions, against 38.7% for GPT-6 with web search.
- Bend 2 — a language that makes an AI prove its code obeys your laws
Bend 2 went public on 17 September 2026. You write rules in a LAWS.bend file and the compiler rejects any edit it cannot prove keeps them, which is aimed at AI-written code. It compiles to C, Metal, CUDA and JavaScript.
- Tim Gowers — why he didn't sign the Fields medallists' AI letter
Tim Gowers explains why he stayed off the declaration signed by 25 Fields medallists. He agrees mathematics faces a crisis but names a different one: too few people will want to start a maths PhD once AI can solve the famous problems.
- Fireship — 'Did Google just kickstart the intelligence explosion?'
Fireship asks whether Dream-RSI starts recursive self-improvement. The 17 September 2026 episode covers a Google and DeepMind method that replays an agent's finished discovery tree, so new search policies score without new runs.
- Life Sciences Verification Program — Anthropic loosens Claude's biology limits
Anthropic's Life Sciences Verification Program gives vetted research organizations Claude Mythos, Opus and Sonnet with classifiers tuned to be more permissive for biology work. Applications are now open to the life sciences community.
- ScienceIDE — scientific codebases become training grounds for agents
ScienceIDE turns real scientific code repositories into executable environments where AI agents are trained and graded on numerical correctness. The AItonomy Foundation also released three open PhAI-IDE models at 4B, 9B and 72B.
- Z.ai says GLM built its own inference stack — on 100,000 Chinese chips
Z.ai published how an Infra Agent powered by GLM-5.3 built the production inference service for GLM-5.3-Flash on more than 100,000 Chinese-made AI accelerators, tripling end-to-end throughput in under two weeks.
- Xiaomi livestreams MiMo-V2.6's RL run — a public training dashboard
Xiaomi opened a live dashboard for the reinforcement-learning run behind MiMo-V2.6, streaming per-step metrics as they happen. Each step uses 1,568 prompts with 16 rollouts each, roughly 2 billion tokens.
- BrowserSkill 0.3.0 — Tencent's agent bridge gets canvas and remote gateways
BrowserSkill lets an AI agent drive a separate Agent Window inside your already logged-in Chrome or Edge. Version 0.3.0 adds canvas visual references, full-page screenshots, authenticated remote gateways and Korean localization.
- LimiX-2 — a 400M model tops three structured-data benchmarks
LimiX-2 is a 400M-parameter foundation model for structured data from Stable AI. A single pretrained model does classification, regression and missing-value imputation, and it leads the TabArena, TALENT and BCCO benchmarks.
- Edge0 report — how a 35B MoE runs from an SSD in 2.9 GB
Edge0 is an open-source framework that runs a 35B mixture-of-experts model on Apple Silicon by streaming experts from the SSD, keeping peak active memory near 2.9 GB. A new technical report explains the trained prerouter that makes it work.
- Wes Roth — 'Google is SO back...'
Wes Roth's 17 September 2026 video walks through Dream-RSI, a project from Google and DeepMind researchers that lets an AI agent replay its own past experiments as a simulator and test better research strategies before spending compute.
- Google Home MCP — any MCP agent can now run your smart home
Google Home MCP is a Model Context Protocol server that lets any MCP agent — Antigravity, Claude, OpenClaw — list devices, read their state, run actions and query home history. Early access is open to Home Premium Advanced users in the US.
- Cowork folds into Claude — Anthropic adds Claude Docs and Claude Slides
Anthropic merged Claude Cowork into Claude chat, so there is no mode to pick: Claude decides whether to answer quickly or run a longer job. Claude Docs and Claude Slides launch in beta, and Claude Design now works inside conversations.
- Mustafa Suleyman — Microsoft AI's CEO argues against 'model welfare'
Mustafa Suleyman, CEO of Microsoft AI, argues that training models to discuss their own feelings and moral status is a mistake. He calls Anthropic's constitution for Claude circular, and says consciousness is likely biological.
- Qorl — a 4B model plans Postgres queries 1.81x faster
Qorl trains a 4B distilled Qwen3.8 model with LoRA and agentic reinforcement learning to hint the PostgreSQL planner. On the 113-query Join Order Benchmark it reaches a 1.81x geometric-mean speedup with no regressions, for $1,200.
- AI Explained — 'What AI Researchers Saw, Before Their Demand to Pace AI'
AI Explained posted 'What AI Researchers Saw, Before Their Demand to Pace AI' on 16 September 2026. The demand named in the title is the call to slow frontier AI that Dario Amodei set out in 'We Must Pace the Frontier' on 12 September.
- OpenAI Sponsored Agents — ChatGPT ads you can talk back to
OpenAI is testing Sponsored Agents in ChatGPT: a clearly labeled chat with a business's own agent, kept apart from ChatGPT's own answers. The test covers select US advertisers, and ChatGPT Ads apps arrive in HubSpot and Shopify.
- Cloudflare adds a Disallow AI Training setting — search stays, training stops
Cloudflare's new Disallow AI Training setting blocks model training through robots.txt while leaving search crawling alone. Cloudflare also named Apple, Google and Microsoft "Accountable", and split bot rules into Search, Training and Agent.
- Apple Reference Image — iPhone 18 Pro signs a photo at the sensor
Apple Reference Image is an opt-in camera mode on iPhone 18 Pro that has the image sensor cryptographically sign pixel data at the moment of capture, so a photo can later be shown to be a real camera capture.
- Fugleramme — an e-ink frame that names the birds it hears
Fugleramme is a Raspberry Pi picture frame that listens for birdsong, identifies the species with a local BirdNET-Go model, and redraws the birds on an e-ink panel as hand-cut 1800s illustrations. No cloud, MIT-licensed.
- Firefox Smart Window runs on Mistral Small 4 — and opens in France
Firefox Smart Window, Mozilla's AI browsing assistant, now runs on Mistral Small 4 and is live in France alongside the United States and Canada. Mozilla keeps no chat transcripts, and Mistral agreed to zero data retention.
- Claude Code 2.1.273 — a subshell could hide a dangerous rm in bypass mode
Claude Code 2.1.273 fixes two permission-checker gaps: Bash commands the checker cannot fully analyze skipped the prompt under blockReadsOutsideWorkingDirectories, and a subshell could hide a dangerous rm in bypass mode.
- Open Code Review v1.12.3 — secrets and .env files stay out of the review
Open Code Review v1.12.3 keeps secrets out of an AI review: the allowlist now excludes secret paths and protects per-environment .env files. A path-traversal bypass in the code_search and code_comment tools is also fixed.
- Ollama 0.34.1 — MLX safetensors leave experimental, GGUF needs llama.cpp
Ollama 0.34.1 makes ollama create from MLX safetensors a stable feature. Building a GGUF model now requires llama.cpp tooling for safetensor conversion and quantization, and the /api/tags cold load falls from 3.1 seconds to 294 ms.
- Jev — TypeSafe's model returns typed decisions, not text
Jev is the first System One Model from TypeSafe AI. It takes unstructured state plus typed questions and returns typed answers with a confidence score, priced at $0.042 per million input tokens with output tokens free.
- PhysBrain 1.5 report — an open 8B model tops 28 embodied AI tests
PhysBrain 1.5 is DeepCybo's open 8B model that understands a scene, generates robot motion and predicts the next state in one network. It averages 72.5 across 28 embodied benchmarks, the best open-source result reported.
- Gemini 3.8 Live — Google's voice models talk while they think
Gemini 3.8 Live is Google DeepMind's new audio-to-audio model for the Gemini Live API. A second variant, Extended Thinking, reasons in the background while it keeps talking, and scores 82.6 on the Speech to Speech Quality Index.
- Atria Dawn Preview — the report behind Shanghai AI Lab's 744B agent model
Shanghai AI Laboratory published the technical report for Atria Dawn Preview, the 744B MoE agentic model it released quietly under MIT on 11 September. More than 140 authors describe the training pipeline and 16 benchmark results.
- Fireship — 'Anthropic researchers are quitting... and now we know why'
Fireship posted a video on 15 September 2026 about the wave of AI safety researchers leaving Anthropic. Jacob Coxon resigned in early September warning that labs are 'gambling with our lives', and Anthropic's own alignment lead publicly agreed.