Latest news, beside the leaderboard

Nearly nine years after its 2017 concept reveal, Tesla hosted a live unveiling of the second-generation Roadster at SpaceX's McGregor rocket test facility in Texas. Using an active underbody fan suction system and SpaceX cold-gas thrusters fed by carbon-fiber COPV tanks, the prototype drove upside down on a curved ramp, executed a mid-air barrel roll recovery, and vaulted over a lineup of Tesla vehicles including a Semi truck. Offered in Performance (<1.6s, $50k deposit), Thrust (<0.9s, $250k deposit), and invite-only single-seat Rocket (<0.59s) trims, production is tentatively targeted for 2028.

On 2026-10-15 Anthropic moved Claude Opus 5.5 fast mode out of research preview and removed the waitlist: no access request and no fast-mode-2026-02-01 beta header needed, just speed: "fast". Same weights, up to 2.5x output tokens/sec per Anthropic, at 2x standard price ($8/$40 per MTok). Start/Build/Scale tiers get a default fast rate limit; it is live on Microsoft Foundry (Azure-hosted) with Google Cloud coming soon. Opus 5 and Opus 4.8 fast mode remain in research preview.

Terminal agents frequently struggle to convert partial command-line progress into complete end-to-end task resolution. Researchers from Tencent AI Lab and NUS introduce Terminal-RSI, alternating Recursive Self-Rewrite (RSR) of verified trajectories with critic-inheriting Recursive Policy-Refine (RPR) alongside Turn-Boundary Skip-Observation GAE (TB-GAE). Evolving Qwen3.8-27B across five stages yields Terminal-RSI-27B, achieving 86.52% on Terminal-Bench 2.1 and 79.00% on Terminal-Bench Hard.

Executing every step of a long-horizon coding trajectory on frontier cloud LLMs incurs prohibitive inference costs. NVIDIA Research introduces EdgeWise, an edge-cloud collaborative framework that deploys a local Small Language Model (SLM) as the primary solver and uses step-level reinforcement learning to selectively delegate only critical steps to a cloud LLM. EdgeWise matches or outperforms pure-cloud LLM execution while cutting inference cost by up to 62% on SWE-Rebench and 34% on Terminal-Bench 2.1.

Standard agent harnesses invoke massive LLMs at every step, even though most steps reduce to discrete multiple-choice decisions such as selecting the next tool. Researchers from UIUC, Yale, and Princeton introduce JevAgentHarness, AgentDecisionBench, and the open AgentDecision-4B model. By calibrating per-choice confidence thresholds and training on post-mistake states, JevAgentHarness makes agents 81.1%–89.0% cheaper and up to 2.5x faster while solving 91.5% of out-of-distribution tasks.

Researchers from USC, University of Rochester, and NVIDIA introduce 3DCodeVerse, an open ecosystem empowering coding agents to synthesize editable and verifiable 3D worlds as code. Featuring the 3DCodeVerse-2M dataset of 2 million standalone executable 3D programs and a multi-agent DAG coding harness with automated spatial measurements and visual verification, 3DCodeVerse achieves a 90% human evaluation win rate against Claude Design (3D Object) when both run on Claude Opus 5.5.

Shanghai AI Lab announced on 2026-10-15 that Disheng Xu (Great Bay University) and Jiesong Zhang (KTH), aided by an internal version of Intern-S2, proved the codimension-one case of Ghys's 1995 holomorphic Anosov conjecture, including complex dimension three (arXiv:2609.37695). Peking University's team formalised all four main theorems in Lean with the SHEAF framework in about five days with up to 100 concurrent agents and ~550k lines, sorry-free and accepted by Lean comparator, with Hsiang–Wall (1969) as an explicit hypothesis.

On 2026-10-12 Meta introduced Muse Wonder and Muse Wonder Full, its most capable models yet for personal and professional agentic work. Official posts highlight Active Memory (carry and retrieve what matters across conversations) and Active Focus (persist through hard tasks, recover from errors, clarify and confirm before consequential actions). Muse Wonder is available today via the Meta Model API and Muse Code; Muse Wonder Full is available today for Muse Code subscribers. Third-party Artificial Analysis Intelligence Index (max reasoning) lists Wonder Full at 56 and Wonder at 52 (Spark 1.3 Max was 48). Update Oct 16: Muse Wonder is now in Cursor with 300k context (1M Max Mode at the same rates), $4/$20 per million tokens, and 46.1% on CursorBench per Cursor.

On Oct 13 Google made Gemini 4 Argon globally available: the stable gemini-4-argon model is live on the Gemini API paid tier and AI Studio, in Antigravity for AI Ultra, and in the Gemini app and AI Mode for Pro and Ultra subscribers. Input, thinking and output share a 1,048,576-token window, and background execution lets one response run up to 1M tokens. Intro pricing is $2 input / $10 output per million tokens through Nov 30, then $4 / $20. There is no free tier, explicit caching is unsupported, and temperature/top_p/top_k are ignored.

On 2026-10-15 Anthropic moved Claude Code’s Code Review out of research preview to general availability for Team and Enterprise organizations. Most reviews finish about 3x faster (roughly 20 minutes down to 6–10). Admins pick Quick, Standard, or Deep per repository; the Claude GitHub App auto-reviews new PRs, or anyone with write access can comment @claude review. Eligible Enterprise orgs can claim at least $10,000 in Code Review credits by January 15, 2027 (credits expire 30 days after claim).

On Oct 15 OpenAI rolled out GPT-Live-1.1 (successor to GPT-Live-1) and GPT-Live-1.1 Quip (successor to GPT-Live-1 mini), full-duplex voice models that become the default voice for Plus-and-up users in ChatGPT Voice, Codex and dots. New voices cover French, German, Hindi, Japanese, Korean and Mexican Spanish, accents are more natural and the constant 'checking...' filler is reduced. Rolling out to all users; API coming soon. The system card reports statistically significant regressions in four synthetic safety categories.

On 2026-10-15 Meta released Muse Spark Open open weights: an 816B-total / 42B-active MoE with a Perception Encoder, natively multimodal (text/image/video/audio in → text out), 1M context, aimed at coding and long-horizon agents. Weights are on Hugging Face (BF16/NVFP4), SageMaker, Model Garden and Foundry; APIs include OpenRouter, Together, Fireworks and Groq. Meta calls it the most capable US open-weight model to date and, with Apple, optimized it to run locally across two M5 Ultra Mac Studios via MLX.

On 2026-10-15 NVIDIA released Nemotron 3.5 Super VL, an open multimodal agent model with 120B total / 12B active parameters: a Mamba-2–Transformer hybrid MoE with a C-RADIO vision encoder, text/image/document/video input, up to 1M context, and built-in MTP plus LiLiCorr speculative decoding. Weights, datasets and recipes ship under OpenMDW-1.1 on Hugging Face, OpenRouter, ModelScope and NIM. Alongside it, NVIDIA published Bolt 74B-A9B 24GB, a natively 2-bit (QAT) local agent model for RTX/DGX Spark; the 4GB Bolt E4B follows on Oct 23.

On 2026-10-15 Black Forest Labs released FLUX 3 Image Klein, a 5.6B open-weights image model for text-to-image, editing and multi-reference generation with up to 10 references, inheriting FLUX 3's native 4K output, pixel-perfect edits and bounding-box control. It ships as a fine-tunable base and a step-distilled model, plus FP8/INT8/NVFP4 builds made with NVIDIA; the INT8 build runs on RTX GPUs with 8GB VRAM. Weights use the FLUX Kommunity License (free commercial use under $5M annual revenue), and it is live on the BFL API.

On 2026-10-15, MBZUAI’s Institute of Foundation Models (IFM) and Inception42 released Jais 3—the first language adaptation of the open K2 Horizon fleet. Six open-weight sizes run from a 0.9B dense phone model to a 375B-A23B sparse MoE (≈23B active), plus Jais Speech for Arabic ASR. Models share one training recipe and agent stack, with up to 512K context (128K on 0.9B), Apache-2.0 weights and stage checkpoints on Hugging Face, and day-0 vLLM support.

Skywork AI (Kunlun Tech) open-sourced Matrix-Game 4.0 on 2026-10-15. An executable game engine owns world state, physics and rules and emits a cheap low-poly RGB proxy; a 5B causal video model built on Wan2.2-Fun-5B-Control re-renders it into finished frames with 1–2 denoising steps. Inference code, a WebRTC browser demo and game source (Apache-2.0) are on GitHub, with Castle, Ancient Egypt and City Racing weights on Hugging Face. Skywork claims 480p/60 FPS and 112 ms end-to-end on an RTX 5090; the repo's own A100 measurements are ~38–39 FPS. Training code and data are not released.

On 2026-10-14 (UTC) ClaudeDevs announced a Claude Code desktop update built for running many sessions side by side: a new-session screen with recent folders, saved environments (several folders, an SSH host or the cloud) with an optional worktree per folder, nested sessions for parallel tasks that report only results back to the main thread, one-click follow-up sessions, /btw side questions in their own tab, and a sidebar that separates working sessions from ones that need you, plus a live to-do list, recap cards and a side rail showing where a session runs, its context and its changes. Update the desktop app to get it; Anthropic published no performance numbers.

On 2026-10-15 StepFun released open weights for Step 5 Preview as promised: a 600B-total / 27B-active sparse MoE with up to 1M-token context and vision input, aimed at agentic software engineering and knowledge work. Checkpoints are on Hugging Face (FP8, BF16, NVFP4) and ModelScope. Official scores include StepCodeBench 49.0%, FrontierFinance 66.4%, and Artificial Analysis Intelligence Index 44. September coverage was API/index only; this is the first public weight drop.

On 2026-10-14 Anthropic put OpenTelemetry tracing for Claude Platform into public beta. Admins add a telemetry destination in the Claude Console and the API exports spans for Messages API requests and Managed Agents sessions server-side with no code changes, while the Claude SDKs export client-side spans for requests and tool-runner tools. Spans follow the OpenTelemetry gen_ai conventions and default to metadata (duration, status, token counts, model, tool names), with an optional full-content mode.

Google Research publicly released Deep Review on 2026-10-14. It grew out of the experimental Paper Assistant Tool (PAT), which gave opt-in pre-submission feedback on 16,000+ manuscripts at STOC 2026, ICML 2026, NeurIPS 2026 and ICLR 2027, and now covers design docs, presentations, and legal and financial documents. Users get two free reviews a week, can connect a Gemini API key for higher limits and the Deep Review API, and the system now supports Gemini 4 Argon.