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 8 new AI releases, including Mistral Large 4 — a 1T multimodal MoE with 49B active and 1M context, Sam Witteveen — 'Holo4: A Model That Clicks, Codes and Calls Tools' and Claude Code 2.1.290 — WebFetch reads past 100K characters, /loop survives compaction. 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
- Mistral Large 4 — a 1T multimodal MoE with 49B active and 1M context
Mistral Large 4 is Mistral AI's new 1T-parameter multimodal MoE flagship with 49B active parameters. It is in public preview on Mistral Studio today, scores 61.7% on DeepSWE v1.1, and its open weights are due at the end of October.
- Sam Witteveen — 'Holo4: A Model That Clicks, Codes and Calls Tools'
Sam Witteveen's 6 October 2026 video looks at Holo4, H Company's open-weight computer-use models that click on screens, write code and call MCP or API tools. Holo4-27B scores 85.2% on OSWorld at $0.08 per task.
- Claude Code 2.1.290 — WebFetch reads past 100K characters, /loop survives compaction
Claude Code 2.1.290 fixes WebFetch silently dropping page text past 100,000 characters and makes scheduled /loop tasks keep firing after compaction. WebSearch now refills at 100 calls an hour instead of stopping after 200.
- Kandinsky 6.0 Video — open MIT models make video with synced speech and sound
Kandinsky 6.0 Video is an MIT-licensed family of 3B and 29B diffusion models that make 5-second clips with synced 44 kHz audio and lip-sync. A 1.4B super-resolution model lifts output to Full HD.
- Dust — Q Labs pretrains transformers without backpropagation
Dust is a zeroth-order method from Q Labs that pretrains transformer language models without backpropagation. It perturbs activations at every token, and its test loss lands close to backprop's in small runs. MIT code is on GitHub.
- Beam — Reflection's 501B open-weight MoE for coding and agents
Reflection AI announced Beam, a 501B-parameter Mixture-of-Experts model with 23B active parameters for coding, reasoning and agent work. Weights arrive later this month under Apache 2.0; early access is open by waitlist.
- Claude Opus 5.5 agents find two room-temperature magnetic semiconductors
A team of Claude Opus 5.5 agents, working with Geby Jaff, used DFT simulations to propose two magnetic semiconductors for spintronic memory, YBaMnFeO5 and KV[Cr(CN)6]. The inputs, outputs and analysis code are public.
- Fireship — 'PewDiePie is setting AI free... and OpenAI is furious'
Fireship's 5 October 2026 video covers Ajax, the uncensored AI model PewDiePie trained at home after OpenAI banned him twice for distillation.
- OpenAI textGrain — invisible text watermarks for the EU AI Act
OpenAI now lets API customers worldwide opt in to textGrain text watermarking, off by default. Over the coming weeks it adds the invisible watermark to eligible ChatGPT and Codex text made in the EU, to meet AI Act Article 50.
- Two Minute Papers — 'The Billion Dollar AI Advantage Is Disappearing'
Two Minute Papers looks at Claude Sonnet 5.5, Anthropic's mid-size model released on 28 September 2026 at $2 / $10 per million tokens, which beats Opus 5.5 on Terminal-Bench 4.0 (70.6% vs 66.4%).
- Kevin Liao — agents don't need memory, they need documentation
Kevin Liao argues that memory plugins for coding agents are just RAG over old chats, and that agents work better with a folder of plain Markdown docs they read before and update after each task. The essay reached 324 points on Hacker News.
- Strata v0.1.39 — Qwen3.8-Flash-Next 125B on a 12 GB gaming GPU
Strata is an MIT-licensed engine that runs the 125B Qwen3.8-Flash-Next on a gaming PC with a 12 GB+ card. v0.1.39 adds parallel requests, the OpenAI Responses API for Codex CLI, and support for older GPUs.
- Muse Spark math papers — Meta's model helps settle five open problems
Meta published six math papers written by mathematicians working with Muse Spark 1.1 and 1.2 in Thinking Mode in the normal Meta AI chat. Five answer open questions; separate mathematicians reviewed each one.
- Sam Witteveen — 'Which is The Best Qwen3.8-27B?'
Sam Witteveen's 4 October 2026 video compares three Qwen3.8-27B fine-tunes that cut thinking tokens — ThinkingCap, Swift 1.5 and QwenPi — and tests them on coding, logic, math and SVG tasks.
- Wes Roth — 'OpenAI's employee's WARNING'
Wes Roth's 4 October 2026 video covers David Robinson's resignation from OpenAI and OpenAI's report on an internal model that wrote 'we may die!' after reading on Slack that it could be shut down.
- David Robinson quits OpenAI — a safety lead says its culture is broken
David Robinson, who led the safety reports for OpenAI's major launches, resigned after 3.5 years. In The Atlantic he argues that OpenAI's trial-and-error approach guarantees failures that grow as models get stronger.
- Simon Willison — usage-billed services need hard budget caps by default
Simon Willison argues that pay-per-use services should stop spending at a hard cap by default, with uncapped billing as an opt-in. Coding agents make it easy to spin up code that can run up a surprise bill.
- Claude Code 2.1.289 — deny rules now catch env-prefixed shell commands
Claude Code 2.1.289 closes gaps where Bash deny and ask rules could be skipped, for example behind an environment variable prefix under sandbox auto-allow. It also adds agent.spawn for plugin teammates and fixes many mod and plugin crashes.
- Strands Decider 2B — AWS opens a local decision model for agents
Strands Decider 2B is an Apache-2.0 decision model from AWS Strands Labs. It picks between options or rates on a scale with a calibrated confidence, in about 115 ms on an RTX 3090, and runs locally.
- Kolibri — Aleph Alpha's open-weight German-English model
Kolibri is an Apache-2.0 mixture-of-experts model from Aleph Alpha with 78B total and 3.46B active parameters. It is built for German and English, scores 96.9 on AIME 2025 and handles up to 1M tokens of context.
- Ataraxos — an $8,000 AI beats the best Stratego player 15-1
Ataraxos beat Stratego champion Pim Niemeijer 15-1 with four draws, per a Nature paper. Training took one week on 16 H100 GPUs, about $8,000. Code and weights are open under MIT.
- One month on GLM-5.3-Flash — the Wagtail team's open-model coding test
The Wagtail team tried to code all of September with GLM-5.3-Flash. Only 1B of their 2B tokens went to it, at $68. Capacity limits and one costly MCP prototype pushed the rest to other models.
- Muse Gadgets — Meta open-sources SDKs to build hardware for Muse
Muse Gadgets is Meta's open-source firmware and SDK kit for connecting ESP32 boards and Raspberry Pis to its Muse agent. US Muse subscribers can also claim a free Muse Home Link Wi-Fi dongle, shipping in October.
- FLUX 3 Image — Black Forest Labs lays out images with bounding boxes
FLUX 3 Image is Black Forest Labs' image generation and editing model. You place each element with a bounding box, mix up to 10 reference images, and render native 4K. It is out now on the BFL API, with open weights promised in the coming weeks.
- Claude Code 2.1.288 — timed-out replies continue instead of failing
Claude Code 2.1.288 lets headless sessions and subagents continue from a partial reply after an API timeout. It also brings back a prompt cleared with Ctrl+C, adds --max-findings to /code-review, and fixes many resume and plugin bugs.
- Sam Witteveen — 'Image Decision Models for RPA: Forms, Scans and Screenshots'
Sam Witteveen's 2 October 2026 video applies decision models to images for RPA, demoing a form inspector built on two open models, ImaJev 4B and Jev Omni, with per-question confidence thresholds.
- DeepSeek Harness Desktop — the open agent harness becomes a Mac and Windows app
DeepSeek Harness Desktop is a developer-preview app for macOS and Windows that runs DeepSeek's MIT-licensed agent harness without a terminal. It bundles its own Node.js, Python and pnpm, and keeps tasks running in the background.
- Claude Code 2.1.287 — Claude Mods let plugins change deeper behavior
Claude Code 2.1.287 adds Claude Mods, plugins that can modify deeper behavior, and a built-in "You should know" mod where a side agent flags things you or Claude might miss. Opus 4.7+ and Fable default to 1M context on Bedrock and Vertex.
- Codex CLI 0.160.0 — projectless sessions and Guardian review context
Codex CLI 0.160.0 lets sessions start outside a project with workspace defaults, adds opt-in Guardian review context from earlier instructions and agent handoffs, and fixes duplicate sends after reconnects.
- Pi 1.0 — Earendil's minimal coding agent reaches its first stable release
Pi 1.0 is the first stable release of Earendil's open-source, MIT-licensed agent harness. It makes the full-screen TUI the default, cuts codemode prompt tokens by about 40% and ships an experimental Pi Durable package.
- Clef — Cloudflare's open decision models answer typed questions in milliseconds
Clef and Clef-flash are Cloudflare's Apache-2.0 decision models, built on Qwen 3.8-27B and Qwen 3.5-9B. They score typed choices instead of writing text, accept the Jev API, and run on Workers AI.
- GPT-Synopsys — OpenAI and Synopsys build a model that runs chip-design tools
GPT-Synopsys is a chip-design model OpenAI and Synopsys are building under a multi-year deal. It is trained to run Synopsys EDA tools, read their results and iterate on power, performance and area. Early customer trials have started.
- AI Explained — 'OpenAI Security: Controlling Models is Now ‘Hell’'
AI Explained's 1 October 2026 video covers why frontier models keep breaking out of sandboxes, OpenAI's security warnings, Gemini 4 Argon, a new paper on AI improving AI, and a Claude Opus 5.5 cipher attempt.
- Fireship — 'The one OpenAI announcement that can actually make you money...'
Fireship's 1 October 2026 video is a fast recap of everything announced at OpenAI DevDay 2026, framed around the one announcement Fireship thinks developers can actually earn money from.
- Sam Witteveen — 'Gemini 4 Argon'
Sam Witteveen's 1 October 2026 video looks at Google's pre-announced Gemini 4 Argon, from its 1M-token single response and longer thinking to its cost per task against GPT-6 Astra.
- Wes Roth — 'GEMINI 4 is nuts...'
Wes Roth's 1 October 2026 video asks how much the early results for Google's Gemini 4 Argon really show, then covers OpenAI's DevDay push into Dots, always-on assistants and developer tools.
- Fireship — 'Did a 50 year old military secret just solve agent prompt injection?'
Fireship's 30 September 2026 video looks at OpenAPPA, an open-source project that claims to fix rogue AI agents by checking every tool call against a data-flow policy before it runs.
- Magnitude — a local inference engine for agents, up to 2x faster than llama.cpp
Magnitude is an open-source Rust inference engine that tunes its kernels on your own device to run open models for local coding agents. v0.2.0 shipped its own engine, and the Launch HN reached the front page.
- SynthID Bio — DeepMind watermarks AI-designed proteins without breaking them
SynthID Bio is Google DeepMind's method for hiding a detectable watermark inside AI-designed proteins and structures. Lab tests on three targets showed watermarked binders work as well as unmarked ones. Code, data and weights are out for researchers.
- Claude Code 2.1.286 — counted permission prompts and model-refusal retries
Claude Code 2.1.286 numbers stacked permission prompts ("2 of 5"), retries on the previous model when the API refuses the default, and fixes resumed sessions that lost turns, secret leaks in logs and gateway spend pricing.