Model intelligence.
Follow the changes. Understand what comes next.
Google Launches Gemini 4 Neon in Antigravity With 1M Context, Four Thinking Levels, and Benchmarks vs. Claude Opus 5.5 and GPT-6.2 Sol
Announced on November 19, 2026 at 05:51 Beijing time, Google's Gemini 4 Neon features a 1M-token context window, four thinking levels, and planner-executor pairing with Gemini 4 Hydrogen in Google Antigravity, outperforming Claude Opus 5.5 and GPT-6.2 Sol across most of 18 published benchmarks.
Latest stories349
Newest publishedAnthropic: Early Progress on Accelerating Drug Development with Claude
Anthropic Science (27 Oct 2026): early progress with Claude across BigHat, BMS, Divergence, Enveda, Generate, and TCG, plus acquiring Mithril Medicine (rare muscular dystrophy). Partner notes include Generate CAR-T design ~70% shorter; Divergence ~35k antibody designs in about a month; BigHat ~10 weeks / ~1,010 variants; Enveda manufacturing docs and filings; TCG’s Pleiades; BMS clinical planning (expand 2027). Anthropic says the work is early and mostly preclinical—not approved medicines. Lab or mouse data have not been verified on this site; for reference only.
Anthropic: Unreleased Claude Model Finds Critical Real-World Vulnerabilities
Anthropic Frontier Red Team (27 Oct 2026): an unreleased internal research model—not a shipping Claude product—found critical real-world flaws. Outcomes: Chrome one-click RCE patched in Chrome 156 (62 critical/high CVEs); Safari SOP bypass (data-only; Apple fixed); Android lock-screen bypass on a stolen locked device (Google addressed); Windows remote DoS CVE-2026-33096 (QUIC); Linux zero-click root CVE-2026-98432 (MPTCP). CVD: 90-day patch + 45 days before discussing beyond public advisories. These chains have not been reproduced on this site; for reference only.
MAI Cyber 1.1 Flash: About 18% Lower Scan Cost Inside MDASH
On October 27, 2026, Microsoft AI’s blog announced MAI Cyber 1.1 Flash inside MDASH. Versus MAI Cyber 1 Flash: comparable performance and 18% less cost per scan on an internal MDASH vulnerability-scanning evaluation suite. Available through MDASH to verified customers (new profile or request access)—not public GA. Separately on X, Mustafa Suleyman said MAI Cyber adoption in the security harness doubled to 50% in 30 days and is 20% more token efficient without sacrificing quality. For reference only.
Claude Code Is Multiplayer Now: Share a Session with Delta (Beta)
On October 27, 2026, Delta’s blog (Franciska Dethlefsen) said Claude Code is now multiplayer in beta via Delta CLI and an open-source Claude Code plugin. /delta share yields a link; teammates use /delta join in their own Claude Code session. The plugin shares the conversation and unfinished code changes / worktree; the CLI syncs files across machines. Each person keeps their own Claude session, with the shared conversation as context. More harness support is coming, starting with Pi. This is a Delta / Zed-side capability, not an Anthropic official Claude Code update. For reference only.
Gemini Robotics ER 2 Preview Lands on Google Cloud and Gemini API
On October 27, 2026, Google Cloud said Gemini Robotics Embodied Reasoning 2 (ER 2) is in public preview on the Gemini platform—a high-level robotics brain for long-horizon planning and agentic tool orchestration. Gemini API IDs: gemini-robotics-er-2-preview and gemini-robotics-er-2-streaming-preview. DeepMind lists public preview on Google AI Studio and the Gemini API; Gemini Enterprise Agent Platform remains private preview. Paid tier is $1/$5 per 1M tokens through December 31, 2026, then $2/$10 from January 1, 2027; Free Tier is free of charge. For reference only.
Google DeepMind Unveils Genie 4 Interactive World Model
Google DeepMind presents Genie 4 as its most realistic and interactive world model yet. Scenes can be reshaped in natural language—move the camera, add characters, change the weather—with native audio. Official limitations say the model can support a few minutes of continuous interaction, not extended sessions. Availability is a research preview for trusted testers, with access coming soon starting with Google AI Ultra subscribers in Project Genie and early access partners of the Gemini API and Google AI Studio. For reference only.
OpenAI and Oracle Bring Oracle Apps and Data into ChatGPT
On October 27, 2026, OpenAI said it is expanding its partnership with Oracle to bring Oracle applications and data into ChatGPT and to power new Oracle products with OpenAI models. Oracle is developing Oracle AI Workspace—coming soon—so approved workflows can run inside ChatGPT. Alongside that, Oracle announced AI Security Mesh, an AI SMB Bundle (ChatGPT Business plus Codex plus recurring OCI credits), and App Composer. U.S. Oracle customers can buy OpenAI API, Codex, and ChatGPT Work on Oracle Marketplace with eligible Universal Credits. For reference only.
Tencent Opens Hy-SimulMT — Download Isn't Instant Interpreting
On October 26, 2026, Tencent Hunyuan released Hy-SimulMT for bidirectional Chinese–English real-time simultaneous translation with real-time voice cloning. Code and weights are public under Apache-2.0; the main checkpoint is Hy-SimulMT-30B-A3B (~30.4B). RealSI reports StreamLAAL of about 3 seconds—a streaming lag metric, not time-to-first-speech. Speech output and cloning need a configured TTS service. A llama.cpp PR (#31198) is still open and unmerged; Hugging Face lists no inference provider. Open weights are not “plug in and interpret.”
Code2Video-AA: How Models Design Videos in Code — and What It Costs
On October 26, 2026, Artificial Analysis published Code2Video-AA—its independent run of HeyGen’s Code2Video Bench. Twenty-seven models each got 114 short product/brand briefs, one call, no tools: HyperFrames HTML (layout + GSAP) rendered to video. Gemini 4 Argon scores pairwise Elo on five axes (anchor GPT-5.5 xhigh = 1000). Claude Opus 5.5 (max with fallback) leads at 1536 Elo (~$3.08/task); GPT-6.1 Sol (max) is fourth at 1369 (~$0.31). This ranks code-designed motion demos—not Kling/Sora-style pixel generators. Dollars are AA evaluation cost per task, not a finished-film quote.
Baidu Open-Sources Unlimited OCR 2 for Long Documents
On October 27, 2026, Baidu open-sourced Unlimited OCR 2 under MIT on Hugging Face and GitHub—about 3.3B parameters. Sparse Reference Sliding Window Attention (SR-SWA) compresses ultra-long text-visual attention into a fixed-size set of working references so per-step latency stays nearly constant as documents grow. Baidu reports SOTA on MPDocBench and MarathonDocBench; those scores are vendor-reported. Multi-page parsing can run in one request via infer_multi, with Transformers for quick tests and SGLang for serving.
GitHub Copilot Assisted Permissions Is Generally Available
GitHub says Assisted permissions is generally available in GitHub Copilot CLI, the GitHub Copilot app, and Visual Studio Code. When Copilot asks to run a command, access a file system path, or open a URL, a large language model evaluates the request: clearly safe ones proceed automatically; everything else still needs manual approval. Enterprise admins can block Allow All while keeping Assisted, or disable Assisted separately, and can set a default permission mode for new sessions.
Devin Security Upgrade Brings Claude Mythos Deep Scans and Enterprise Credits
On October 26, 2026, Cognition announced a major upgrade to Devin Security and a commitment of hundreds of millions of dollars in Devin credits. Deep scans are powered by Claude Mythos and other frontier cyber models; enterprise teams use them through Devin without separate model access. Devin Security builds a living custom threat model with the team, re-scores findings by exploitability and potential damage, and can write, validate, and end-to-end test fixes on real VMs in thousands of parallel sessions. Credits cover free frontier scans and remediation. As of October 27, 2026.