AI Agent2026
Runs an external, reviewable coding-agent harness that turns intent into repeatable software work: clarifies requirements, builds reviewed plans, executes in tmux-backed sessions, and collects durable verification. Ships Telegram/Discord delivery, a research REPL, and optional desktop-control tools; beta-stage.
Combines internalizing general skills with task-specific skill utilization via a difficulty-aware router to improve in-distribution and out-of-distribution performance for agentic RL. Uses privileged distillation for hard tasks and diagnostic probing for easy tasks; evaluated on ALFWorld and WebShop.
AI Agent Papers2026
Proposes TASTE, an automatic pipeline that synthesizes challenging agent benchmark tasks by sampling and evolving valid tool-sequence patterns; uses an adaptive contrastive n-gram model and LLM validity judgments to produce τ^c-Bench with broader tool-use coverage and higher difficulty.
AI Agent2026
Local-first AI agent workspace for authoring, running, and recovering agent executions — records model messages, tool calls, tool results, permission decisions, and termination events in an append-only Runtime Event Log. Provides Desktop (Electron), TUI/CLI, and headless Eval surfaces plus local tools and runtime features for pruning, compaction, and durable recovery.
AI Model2026
Hybrid LFM2.5 text-generation model optimized for on-device assistants and agentic workflows — 8.3B total / 1.5B active parameters with 131,072-token context. Prioritizes low-latency, high-throughput inference and multilingual instruction-following; not optimized for pure heavy programming or knowledge-heavy QA without retrieval.
AI Model2026
GGUF quantizations of Step-3.7-Flash: a sparse multimodal Mixture-of-Experts LLM with native image understanding, selectable reasoning levels, and a 256K context window. Ships multiple calibrated Q3/Q4/IQ quant files plus an mmproj vision projector for local llama.cpp inference on high-memory hosts.
AI Agent Papers2026
Analyzes when masking stale observations improves long-horizon search agents and why, identifying an asymmetric inverted-U relationship between masking benefit, retriever quality, and model capacity; explains a token-for-turn trade-off and releases evaluation scaffolds and trajectories.
AI Agent Papers2026
Automates distillation of heterogeneous traces from a target person or role into versioned, inspectable skill packages for LLM agents — producing separate capability and bounded-behavior tracks that support natural-language corrections, rollback, and cross-host installation. Ships with an open system and a skills gallery.
AI Agent Papers2026
Localizes harmful span-level errors inside long research-agent trajectories to show which trajectory segments make final answers unreliable. Provides a 1,000-instance TELBench of annotated spans and DRIFT, a claim-centric auditing method that improves span-level localization and first-error accuracy by up to 30 percentage points.
AI Dataset2026
Provides the gated, official OSWorld 2.0 Python task class files (task_*.py) required to run the benchmark; distributed via a Hugging Face gated dataset to reduce benchmark leakage. Download requires accepting gated access on Hugging Face.
A 20B retrieval subagent trained with reinforcement learning inside a stateful search harness that externalizes recoverable search state (candidate pool, curated evidence, verification records). The harness lets the policy focus on semantic search decisions, improving curated recall and transfer robustness.
AI Agent Papers2026
Workflow-aware benchmark for autonomous medical-AI research that splits agent execution into five stages (Plan, Setup, Validate, Inference, Submit) and evaluates long-horizon runs across segmentation, image enhancement, VQA, report generation, and lesion detection with stage-level scoring.