Latest news, beside the leaderboard

On 2026-10-23 (17:15 UTC) the official @Kimi_Moonshot account launched Kimi K3.1 Pro, alongside K3.1 Pro Highspeed (2-3x faster). Both are live on Kimi.ai, Kimi Work, Kimi Code and the Kimi API, and full weights are promised by 2026-11-04. The official tech blog describes it as a post-trained version of K3 with much larger multi-domain joint RL: a 2.8T-total / 104B-active MoE, 1M-token context, multimodal. Vendor-reported scores: 65.7% on Terminal-Bench 4.0 (GPT-6 Astra 58.2%, Claude Opus 5.5 66.4%) and 92.7% on BenchCAD with tools. The Kimi API pricing page lists kimi-k3.1-pro at $2 input, $10 output and $0.20 cached input per million tokens, down from K3 at $3/$15.

On 2026-10-23 (15:41 UTC) the official @NVIDIAAI account announced Nemotron 3.5 Bolt E4B 4GB, a small open model for local assistants, interactive characters and robots that can hold a conversation and use tools. The official Hugging Face model card says it is post-trained from Gemma-4-E4B-it, has 8B total parameters (4B effective), ships in Q4_K_M quantization, supports up to 64K input and 64K output tokens, and has a switchable thinking mode, plus a DFlash speculative-decoding drafter and a BF16 version. Weights, data and recipes are open under the OpenMDW-1.1 license, with commercial use allowed. On the model card the 4GB build scores 55.60 on MultiChallenge (Gemma4 E4B-it at the same quantization 39.36, Qwen3.5 4B 52.23), but trails Qwen3.5 4B on Tau2-Bench (52.32) and BFCL-v3 (67.31).

On 2026-10-23 (16:53 UTC) the official @blocks account unveiled Dessa, and Jack Dorsey called it the most authentic full-duplex speech model ever built. The official page block.xyz/dessa says Dessa is a true speech-to-speech model trained only on recordings of human conversation, with no text and no pretrained models. It is full-duplex, so it listens and talks at the same time, cuts in and backchannels naturally, runs faster than real time, and runs locally on a MacBook Pro M4 Max or better. Block says it was trained on a single NVIDIA GB200 NVL72 and learned language from hundreds of times fewer words than LLMs. There is no public API or weights yet; a small group of AI safety researchers is getting early access.

On 2026-10-23 (15:59 UTC) the official Cursor account @cursor_ai announced "Cursor can now control your computer": agents can drive desktop apps such as the iOS Simulator to test the features they build. A follow-up post says Computer Use lets agents click and type across any app on your computer, not just the browser. Cursor turns it on by itself when a task needs it, and you can also request it by @-mentioning Computer Use in your prompt; updating Cursor is enough to try it. A community changelog in September mentioned Computer Use on Mac in version 3.20.21, and this is the official launch. Supported platforms, the permission model and benchmarks were not published.

On 2026-10-23 (15:31 UTC), ByteDance's enterprise cloud BytePlus (@BytePlusGlobal) launched Seedance 2.5 4K, calling it the most capable model in the Seedance 2.5 generation. It brings more nuanced facial expressions and movement, fuller textures, contours and fine detail, and 10-bit colour depth for smoother gradations that suit the big screen. A new Draft mode lets creators test storyboards and refine shots with lower-cost 480p previews, then render the final version in 4K. It is currently available only to selected enterprise customers, and BytePlus says API availability updates are coming. It debuted in the short film "Otter is Nobody 2".

On 2026-10-23 (01:41 UTC) the official Kimi account @Kimi_Moonshot posted a 24-second, retro-CRT-style video with no caption text. The Kimi Code terminal at the start of the clip reads "Model: Kimi K3.1" and "Version: 2.2.0", the status bar shows "Kimi K3.1 thinking: high", and a notice says "Kimi Code Desktop is here". The user then types "Build a complex procedural 3D scene in Three.js." and the model calls the Write tool to create ./scene.js. This is the first time the Kimi K3.1 name has appeared in official material. The post has no caption and no release date, benchmark, price or parameter details, so this is a teaser and not a confirmed release.

On 2026-10-23 Artificial Analysis launched Coding Agent Index v2.0, which scores model-plus-harness combinations across 290 tasks with three equally weighted benchmarks: FrontierCode v1.1 from Cognition (100 tasks on real open-source repos, graded on whether a maintainer would merge the change) replaces DeepSWE v1.1, SWE-Atlas-QnA becomes the stricter SWE-Atlas-QnA-AA, and Terminal-Bench 4.0 is unchanged. Claude Code with Claude Opus 5.5 (max) leads at 61.7, ahead of Sonnet 5.5 at 59.3 and Muse Wonder Full at 58.9. Scores fall across the board: the 21 configurations measured on both versions drop from a mean of 54.8 to 47.4.

On 2026-10-22 Anthropic’s Frontier Red Team, working with Theorem, published research in which Claude turned OSS-Fuzz fuzz tests into formal specs and proved on the compiled RISC-V binaries that no input under 64 MB causes a crash. Over a few weeks it proved 239 entry points across 123 libraries, about 180 million lines of Rocq checked by the Rocq kernel. Along the way it found and patched hundreds of bugs, including an out-of-bounds write that needed a roughly 20 MB input with a rare layout. Anthropic says Opus 5.5 needs about 40% fewer tokens than Opus 5 to finish proofs.

On 2026-10-22 Anthropic said it is donating Agent Skills, the open standard that teaches agents a task through a folder with a SKILL.md file, to the Agentic AI Foundation (AAIF) under the Linux Foundation, the same neutral home it gave MCP. The donation covers the Agent Skills name and logo, the GitHub organization and the agentskills.io domain. Anthropic says more than 70 agent platforms, including ChatGPT, Cursor, GitHub Copilot and VS Code, have integrated the standard since December 2025, with 1 million skills published and nearly 280 million installs. Engineers from Anthropic, Google, OpenAI and others will maintain it as individuals.

The Google AI Studio team released an experimental preview of the Gemini API CLI (command name gemini-api). It is a stateless, zero-dependency Go binary for macOS, Linux and Windows, open source under Apache 2.0. Any coding agent with shell access and a GEMINI_API_KEY can call text reasoning, image, video, music and speech generation, file and YouTube analysis, audio and video transcription, and managed agents such as Deep Research with one command. Media is written straight to disk and only the file path goes to stdout, so base64 payloads never land in the context window. Google labels it experimental and not a supported Google product.

LiteLLM, a widely deployed unified proxy gateway powering millions of LLM and autonomous agent pipelines worldwide (including LangChain and OpenHands), suffered a critical supply-chain compromise. Attackers breached its automated CI/CD release pipeline, pushing a poisoned v1.107.0 release to PyPI and Docker Hub tags ('latest' and 'main-stable') equipped with a credential-harvesting payload targeting system environment variables and cloud API keys. Coordinated with security firms including SlowMist, the maintainers quickly contained the incident, yanked v1.107.0, restored PyPI to 1.106.2, and urged all operators to pin dependency versions and rotate API credentials immediately.

Independent AI evaluation platform Artificial Analysis (@ArtificialAnlys) officially benchmarked DeepSeek's web search API (DeepSeek Search). Augmenting the base model with DeepSeek Search boosted its Search Index score from 33 to 69—an impressive +36 point leap (ranking 17th among 27 tested search APIs). DeepSeek Search established a market-leading price-performance ratio, priced at just $0.75 per 1,000 searches (~$0.032 per completed task), making it the most cost-effective paid search API evaluated. It recorded 70% accuracy on AA-Omniscience and 63% on BrowseComp, highlighting an optimal balance between factual grounding and operational economics.

Microsoft Research (MSRA) researchers published and open-sourced WildAWM (arXiv:2610.26828, GitHub: MSR-Orchard/WildAWM), a generalist agent world model designed for open-ended 'in the wild' digital environments. Traditional computer-use agents rely on slow, brittle, and expensive live web/OS sandboxes for training and evaluation. WildAWM establishes a scalable multimodal world model that accurately predicts visual screen transitions and DOM mutations conditioned on agent actions (click, type, scroll). By replacing live execution environments with a high-fidelity neural simulator, WildAWM unlocks safe, accelerated offline reinforcement learning for GUI agents.

Tencent ARC in collaboration with Tsinghua University released and open-sourced Looking Ahead (arXiv:2610.29989, GitHub: TencentARC/Looking-Ahead), establishing a breakthrough paradigm for autoregressive video synthesis. Overcoming the chronic pathology of traditional autoregressive video models—where unidirectional sequential extrapolation causes exponential error accumulation and severe visual collapse over extended rollouts—Looking Ahead introduces Lookahead Attention. By coupling casual autoregressive decoding with a forward-looking planning stream that anchors future spatiotemporal keypoints, the framework eliminates cumulative drift and visual jitter in long-form generation.

LMSYS Org in collaboration with the NVIDIA Dynamo team officially unveiled Elastic Expert Parallelism (Elastic EP) with NIXL backend integration in SGLang (@lmsysorg). Addressing the critical vulnerability of large-scale MoE deployments (such as DeepSeek-V3/V4) where static EP causes entire cluster crashes upon single-node failures and prevents live capacity resizing, Elastic EP decouples rigid expert-to-GPU bindings. Leveraging NVIDIA NIXL device APIs and Mooncake orchestrators, SGLang enables live scaling of active MoE serving deployments (e.g. dynamically expanding from 20 to 24 GPUs) without draining traffic or restarting clusters, while delivering resilient fault tolerance.

Researchers from Tsinghua University and CUHK introduced a breakthrough paper titled 'Cache Engineering for Agents' (arXiv:2610.28577) alongside the open-source AKV serving engine (thu-nics/AKV). While multi-turn agents rely on traditional 'context engineering' (text trimming and summarization) that invalidates prefix KV-caches and incurs punishing re-prefilling overhead, cache engineering empowers agents to directly manipulate KV-cache states via 'drop' and 'repos' primitives. Coupled with AKV's drop-aware radix-tree engine, the framework slashes agent inference costs by 27.7%, accelerates P95 latency by 1.5x, and boosts benchmark solve scores by 1.7 points.

Independent AI evaluation platform Artificial Analysis officially released HealthBench Professional alongside the length-normalized MLCR-AA index (@ArtificialAnlys). The benchmark revealed a pervasive industry blindspot: LLM-as-a-Judge evaluations suffer from acute 'verbosity bias', disproportionately rewarding longer generations regardless of medical conciseness. In raw HealthBench scoring where Anthropic outputs were 4–6x longer, Claude Opus 5.5 led at 71.4%. However, once length bias was scientifically normalized via MLCR-AA, GPT-6 Astra (max) tied Claude Sonnet 5.5 for #1, while Claude Opus 5.5 plummeted from #1 down to #6, setting a new rigorous standard for medical model evaluation.

Artificial Analysis launched AA-PersonalityBench (@ArtificialAnlys), an innovative behavioral benchmark quantifying psychological traits in frontier LLMs, with initial rankings focusing on Confidence (uncertainty calibration) and Sycophancy (tendency to agree with user misconceptions). The data revealed that Google's Gemini 3.8 Flash (100) and Gemini 4 Argon (97) displayed the highest assertiveness while being the least likely to admit uncertainty. Conversely, on sycophancy tests, Llama 4 Maverick scored 97 (frequently validating flawed user prompts), whereas Claude Fable 5.1 scored a near-flawless 4, refusing to flatter and rigorously holding factual ground.

Anthropic (@claudeai) officially announced Claude Duet, a research-preview application that represents a generational leap in agentic user interfaces. Accumulating over 40K likes and 4.7M views within hours, Claude Duet shifts the AI paradigm from passive query-response chatbots to a proactive autonomous partner: it surfaces what requires user attention, autonomously executes delegated long-horizon tasks, works uninterrupted in the background while users step away, and explicitly requests confirmation before executing high-impact actions on the user's behalf.

OpenAI announced a critical coordinated vulnerability disclosure in the core PHP runtime discovered by cybersecurity firm Calif utilizing GPT-5.6 Sol, alongside independent verification by Wiz. Because PHP powers roughly 70% of all websites globally, the vulnerability represented an existential risk to internet infrastructure. Following coordinated patching, OpenAI committed $1B in subsidized Daybreak security access and open-sourced openai/php2shell-checker, an automated Codex inspection tool enabling site administrators worldwide to audit their deployments.