FlavourBench is an automated benchmark that evaluates language models on culinary tasks by providing a versioned, executable ground truth system. Each task presents eight ingredients and asks the model to propose a three‑ingredient portfolio, with all 56 possible portfolios scored by the Epicure system before model execution. The benchmark assesses 27 frontier endpoints across 534 tasks, yielding a FlavourBench Score that averages frozen task scores across families, and includes extensive statistical analysis and reproducible data releases.
By Josef Chen, Erim Hayretci
arXiv:2609.06107v1 Announce Type: new
Abstract: Data policies for reinforcement learning with verifiable rewards (RLVR) determine which rollouts are used, how strongly they are weighted, and which do...
By Hao Liang, Mingrui Chen, Hengyi Feng, Meiyi Qiang, Wentao Zhang
The paper reports that applying reinforcement learning (RL) to the Kimi K2.7 Code model on 1,700 agentic coding tasks improves its performance on six external benchmarks. After a single epoch of GSPO training on a rank‑32 LoRA adapter, pass‑@1 scores increased across all benchmarks, with significant gains even on data released after training. The trained model also reduces agent steps and avoids common failure modes such as dropping requirements or breaking existing behavior.
By Sushant Mehta, Logan Ritchie, Edwin Chen
arXiv:2609.21208v1 Announce Type: new
Abstract: Self-play methods that co-train a single language model as both coder and test author promise to move code-generation RL beyond fixed test suites, but...
By Ana Nunez, Peyman Najafirad
UpgradeBench is a decision‑centric longitudinal benchmark that evaluates how fine‑tuned language‑model specialists should be handled when new base‑model releases occur. It covers four consecutive Qwen releases, a continuation checkpoint, six tasks, two model sizes, and OLMo checkpoints with known training lineage, and examines whether retraining, adapter transfer, or other recovery strategies improve specialist performance. The benchmark reveals that upgrade gains vary by task and release interval, that direct adapter copying is sensitive to pretraining distance, and that teacher relabeling can recover specialists without new annotations.
"whyItMatters":"The study provides actionable insights into the cost‑effective management of specialist models across model releases, showing how to balance retraining effort with performance gains."
By Ye Chen, Weining Zhang
The paper reports on training and calibration experiments for three open models on the ARC‑AGI‑2 benchmark, using offline runs on four L4 GPUs for up to 12 hours. For the NVARC Qwen3‑4B grid model, a curated corpus of 12,000 executable rules and 95,993 tasks was built, and fine‑tuning on 1,920 real task records raised the semi‑private score from 28.96 to 33.06, with no synthetic‑task checkpoint outperforming the real‑task model. For the Qwen3.8‑27B reasoning model, fine‑tuning on 1,036 of its own reasoning traces and a specialized inference procedure increased the score from 26.81 to 44.03, while the Qwen3.8‑Flash‑Next mixture‑of‑experts model achieved a best score of 56.67 after expert pruning, routing statistics adaptation, and an FP8 indexer cache.
By Nguyen Minh Thang, Trinh Thi Phuong
arXiv:2608. 05466v1 Announce Type: new Abstract: High-quality long-horizon training data for terminal agents is expensive to produce, often costing hundreds to thousands of dollars per task, because each task must keep the instruction, environment, reference solution, and verifier mutually consistent.
By Zhongzhi Li, Yucheng Shi, Zongxia Li, Ruhan Wang, Anhao Li, Zixun Huang, Junyao Yang, Lei Ke, Ninghao Liu, Haitao Mi, Leowei Liang
StoreBench is a live‑commerce environment where an autonomous agent runs a mid‑size online apparel store on a production‑grade backend, facing continuous customer orders, supplier price changes, and market shocks. The agent uses the same 29 merchant tools as a human operator, with a windowed operation budget that ties simulated time to actions, ensuring that model latency does not affect the simulation. Seven large language models were evaluated across 11 scenarios of 30–45 days and a full simulated year, but none matched the scripted smart‑triage policy, and human experts outperformed all models.
By Daksh Raghuvanshi, Ved Vedere, Yifan Wang
The paper introduces an open midtraining recipe that primes base models for agentic post‑training, filling a gap between pretraining and full fine‑tuning. It trains on 256K context with hundreds of billions of tokens drawn from natural software‑engineering data and a small share of agentic traces. Applying this midtraining to Nemotron 3.5 Lightning and Super models yields measurable gains on SWE‑bench and DeepSWE benchmarks after identical full SFT, and improves robustness across agent harnesses.
By Soumye Singhal, Joseph Jennings, Ka Shun Shum, Alexander Bukharin, Khushi Bhardwaj, Zirui Liu, Venkat Srinivasan, Jonathan Li, Yash Jain, Kezhi Kong, Markus Kliegl, Olivier Delalleau, Sanjeev Satheesh, Mostofa Patwary
The paper reports that in English all‑words word sense disambiguation (WSD), the scarcity of high‑quality labels—not the models—has become the limiting factor. The authors introduce lexEN, a human‑adjudicated correction layer over the Maru2022 ALL_NEW benchmark, and SenseBench, a living leaderboard for LLM WSD evaluation. They show that frontier large language models reach about 95 % accuracy on lexEN‑v1, that relabeling corpora with these models improves downstream systems, and that fine‑grained WordNet senses are often ill‑posed, with coarsening improving both annotator agreement and model performance.
"whyItMatters":"The study highlights that improving label quality and managing annotation costs are now the critical challenges for advancing WSD performance, as model accuracy is already near its theoretical ceiling."
By Vassili Philippov, Amro Salman, Dmitrii Andreev, Penny Hands, Emil Kaiumov, Pavel Katunin, Anton Nikolaev
arXiv:2606. 01080v1 Announce Type: cross Abstract: Large language models often improve on difficult tasks by spending inference-time compute on a reasoning trace before producing the final answer.
By Dhruv Saini, Rohan Pandey
The paper evaluates a manager‑worker scaffold that uses a shared filesystem workspace to orchestrate multi‑agent large language model (LLM) coding tasks without training or tuning. Across nine models—including five open‑weight and four closed‑weight systems—the scaffold yields statistically significant accuracy gains for some models (e.g., Qwen3.8‑27B, GPT‑5.6‑Luna, GPT‑5.6‑Terra, Kimi‑K3, Minimax‑M3) while producing null or negative effects for others (e.g., Qwen3.6‑35B). The study shows that the manager can triple token usage but still achieves higher accuracy at a fraction of the cost compared to larger single‑pass models, with key mechanisms identified as context management and problem decomposition.
By Victor Gao (Sang Won), Vida Khosrowshahi (Sang Won), Ali Khosrowshahi (Sang Won), Xihao Sun (Sang Won), Juhyun Lee (Sang Won), Simon (Sang Won), Lee