Research Papers Kimi K3 is an open Mixture-of-Experts model with 2.8T parameters (104B activated per token), native vision input and a 1-million-token context window, built on Kimi Delta Attention and post-trained with reinforcement learning across general, agentic and coding tasks.
LingBot-World 2.0 upgrades the game world model to an unbounded interaction horizon with consistent output quality via a causal pretraining paradigm, distills a real-time variant fast enough to drive 720p video at 60 fps, and greatly expands the action set (attacking, archery, spell-casting, shooting) plus text-driven interactive elements.
FreeToken rethinks local inference around two realities of edge AI: agent workloads keep changing shape, and every machine's resource balance differs. It co-designs model layout, expert residency, CPU-GPU execution and memory management instead of committing to a fixed offloading strategy.
VoiceMem gives duplex speech models a working memory that keeps facts and emotional context across a conversation, so a voice assistant can remember both what you said and how you said it, in real time.
Prime Agent is an open harness for long-horizon coding agents built on the Recursive Language Model idea: a persistent IPython session does test-time computation, while histories, memories, skills and subagent specs carry over between runs.
Inspired by human working memory while copying, this technical report builds an end-to-end OCR model whose decoding cost stays flat as output text grows long, avoiding the KV-cache slowdown and memory creep of LLM-decoder OCR systems.
This technical report upgrades LingBot-VLA with a revamped data pipeline of about 60,000 pretraining hours: 50K robot-trajectory hours across 20 embodiments plus 10K egocentric human-video hours. It extends control beyond dual arms to heads, waists, mobile bases and dexterous hands, aiming squarely at the gap between lab robotics and real deployment.
RoboDojo is a unified sim-and-real benchmark for generalist robot manipulation policies, with 42 simulation tasks and 18 real-world tasks probing generalization, memory, precision and long-horizon control. It ships parallel Isaac Sim evaluation plus a cloud-accessible real-eval system and a leaderboard the authors populated with 30 policies.
Apodex 1.1 targets what the authors call working capability: it scales up the diversity and verifiability of executable file, search and code environments, while training agents to decompose long-horizon tasks, delegate parallel work and integrate asynchronous results.
Alaya-EVOKE externalizes persistent scene geometry into a camera-indexed world state bank so the denoiser context stays bounded, and trains with a sparse-attention teacher that supervises long horizons at linear cost, exposing content drift and supporting prompt changes mid-sequence. The teacher distills into a three-step student that the authors report producing each 1.5-second chunk in 2.11s on a single H200 GPU.
Dream-RSI is a framework for letting coding agents improve their own exploration strategy instead of keeping a fixed one. It builds a replay simulator from the agent's past discovery trees, lets the exploration policy practice inside that simulator for cheap off-policy feedback, then redeploys the refined strategy online so new discoveries feed the next round. The authors report competitive or better discovery quality on algorithm engineering, math optimization, and GPU kernel tasks, with substantially lower discovery cost in several settings.
Xiaomi's robotics team presents a vision-language-action foundation model for mobile manipulation, pretrained on more than 100K hours of real-world UMI trajectories with auto-labeled scene-transition language, then post-trained to match embodiments and human imperative prompts. The paper claims clean scaling with more data and parameters, out-of-the-box performance in unseen environments and efficient fine-tuning for dexterous tasks.
Atria Dawn introduces a foundation agentic language model built for scientific research and engineering workflows. Its training pipeline, which the authors call the Verifiable Experience Pipeline, ties tool-mediated interactions to executable environments and externally checked outcomes; the paper also studies human-AI collaboration using 769 task records from 56 researchers working with the agent.
This harness maintains an explicit task state outside the agent's execution context and updates it only with facts independently verified from the environment, using a Manage-Execute-Audit loop so mistaken self-assessments do not snowball across many steps.
Frontis-MA1 ships with OpenMLE, an open full-stack system (task gyms, RL operator learning, evolutionary search) for studying recursive self-improvement. Its 35B meta-evolution agent, post-trained around Draft, Improve, Debug and Crossover program-evolution operators, claims MLE-Bench Lite results the authors say exceed GPT-5.5 plus Codex, with components transferring to held-out NatureBench Lite.
NeoHorse-1 records, for every user turn, the predicted capability demand, the selected service tier and how the interaction went, then converts these records into training examples admitted through structural validation and multi-dimensional semantic checks, feeding them back as the next round of agentic post-training.
ABot-World-0 is an action-conditioned video world model trained on multi-source data from AAA games, simulation engines and internet videos, controlled through raw keyboard input with reference-character memory for consistent third-person rollouts. The authors report streaming 720p at up to 16 FPS on one RTX 5090 with 1.2s action-to-first-frame latency, targeting real-time long-horizon closed-loop interaction.
ZGCM-1 is a fully open 7B dense foundation model aimed at math reasoning and agentic search. The idea is that a smaller model cannot memorize the whole web, so it couples long internal thinking with active tool use, using interleaved gated sliding-window and full attention, an FP8 Muon optimizer, and curriculum training that scales context to 256K. The authors say they release stage-wise weights, training code, per-stage data recipes, and W&B logs.
AuK unifies speech generation and editing in a single open-source model: you describe the desired change in natural language, and it either synthesizes new speech or modifies existing audio accordingly.
OSWorld 2.0 from the xlang-ai team replaces short scripted tasks with 108 long-horizon computer-use workflows built on authentic artifacts and stateful user profiles; humans take a median of about 1.6 hours per task. The headline finding is how far agents still are: the best model, Claude Opus 4.8, completes only 20.6% of tasks, often losing track of constraints or skipping verification.