FrontierFinance: A Long-Horizon Computer-Use Benchmark of Real-World Financial Tasks

As concerns surrounding AI-driven labor displacement intensify in knowledge-intensive sectors, existing benchmarks fail to measure performance on tasks that define practical professional expertise. Finance, in particular, has been identified as a domain with high AI exposure risk, yet lacks robust benchmarks to track real-world developments. This gap is compounded by the absence of clear accountability mechanisms in current Large Language Model (LLM) deployments. To address this, we introduce FrontierFinance, a long-horizon benchmark of 25 complex financial modeling tasks across five core finance models, requiring an average of over 18 hours of skilled human labor per task to complete. Developed with financial professionals, the benchmark reflects industry-standard financial modeling workflows and is paired with detailed rubrics for structured evaluation. We engage human experts to define the tasks, create rubrics, grade LLMs, and perform the tasks themselves as human baselines. We demonstrate that our human experts both receive higher scores on average, and are more likely to provide client-ready outputs than current state-of-the-art systems.

Faster Superword Tokenization

Byte Pair Encoding (BPE) is a widely used tokenization algorithm, whose tokens cannot extend across pre-tokenization boundaries, functionally limiting it to representing at most full words. The BoundlessBPE and SuperBPE algorithms extend and improve BPE by relaxing this limitation and allowing the formation of superwords, which are combinations of pretokens that form phrases. However, previous implementations were impractical to train: for example, BoundlessBPE took 4.7 CPU days to train on 1GB of data. We show that supermerge candidates, two or more consecutive pretokens eligible to form a supermerge, can be aggregated by frequency much like regular pretokens. This avoids keeping full documents in memory, as the original implementations of BoundlessBPE and SuperBPE required, leading to a significant training speedup. We present a two-phase formulation of BoundlessBPE that separates first-phase learning of regular merges from second-phase learning of supermerges, producing identical results to the original implementation. We also show a near-equivalence between two-phase BoundlessBPE and SuperBPE, with the difference being that a manually selected hyperparameter used in SuperBPE can be automatically determined in the second phase of BoundlessBPE. These changes enable a much faster implementation, allowing training on that same 1GB of data in 603 and 593 seconds for BoundlessBPE and SuperBPE, respectively, a more than 600x increase in speed. For each of BoundlessBPE, SuperBPE, and BPE, we open-source both a reference Python implementation and a fast Rust implementation.

Cost-Efficient Estimation of General Abilities Across Benchmarks

Thousands of diverse benchmarks have been developed to measure the quality of large language models (LLMs). Yet prior work has demonstrated that LLM performance is often sufficiently explained by a small set of latent factors, or abilities. This suggests the potential for more efficient and principled benchmarking, but it remains difficult to compare the quality of different methods. Motivated by predictive validity, we argue that the quality of a benchmarking framework should be grounded in how efficiently it enables the prediction of model performance on unseen tasks. To analyze this objective, we collect the "Wide-scale Item Level Dataset" (WILD), a dataset of item-model response pairs, comprising evaluations of 65 models on 109,564 unique items spanning 163 tasks drawn from 27 datasets. This dataset enables the first analysis of how different techniques can predict a model's performance on a large, diverse collection of unseen tasks under different budget constraints. We demonstrate that combining a modified multidimensional item response theory (IRT) model with adaptive item selection driven by optimal experimental design can predict performance on 112 held-out benchmark tasks with a mean absolute error (MAE) of less than 7%, and can do so after observing only 16 items. We further demonstrate that incorporating cost-aware discount factors into our selection criteria can reduce the total tokens needed to reach 7% MAE from 141,000 tokens to only 22,000, an 85% reduction in evaluation cost.

The Effect of Scripts and Formats on LLM Numeracy

Large language models (LLMs) have achieved impressive proficiency in basic arithmetic, rivaling human-level performance on standard numerical tasks. However, little attention has been given to how these models perform when numerical expressions deviate from the prevailing conventions present in their training corpora. In this work, we investigate numerical reasoning across a wide range of numeral scripts and formats. We show that LLM accuracy drops substantially when numerical inputs are rendered in underrepresented scripts or formats, despite the underlying mathematical reasoning being identical. We further demonstrate that targeted prompting strategies, such as few-shot prompting and explicit numeral mapping, can greatly narrow this gap. Our findings highlight an overlooked challenge in multilingual numerical reasoning and provide actionable insights for working with LLMs to reliably interpret, manipulate, and generate numbers across diverse numeral scripts and formatting styles.

On Finding Inconsistencies in Documents

Professionals in academia, law, and finance audit their documents because inconsistencies can result in monetary, reputational, and scientific costs. Language models (LMs) have the potential to dramatically speed up this auditing process. To understand their abilities, we introduce a benchmark, FIND (Finding INconsistencies in Documents), where each example is a document with an inconsistency inserted manually by a domain expert. Despite the documents being long, technical, and complex, the best-performing model (gpt-5) recovered 64% of the inserted inconsistencies. Surprisingly, gpt-5 also found undiscovered inconsistencies present in the original documents. For example, on 50 arXiv papers, we judged 136 out of 196 of the model's suggestions to be legitimate inconsistencies missed by the original authors. However, despite these findings, even the best models miss almost half of the inconsistencies in FIND, demonstrating that inconsistency detection is still a challenging task.

BLEUBERI: BLEU is a surprisingly effective reward for instruction following

Reward models are central to aligning LLMs with human preferences, but they are costly to train, requiring large-scale human-labeled preference data and powerful pretrained LLM backbones. Meanwhile, the increasing availability of high-quality synthetic instruction-following datasets raises the question: can simpler, reference-based metrics serve as viable alternatives to reward models during RL-based alignment? In this paper, we show first that BLEU, a basic string-matching metric, surprisingly matches strong reward models in agreement with human preferences on general instruction-following datasets. Based on this insight, we develop BLEUBERI, a method that first identifies challenging instructions and then applies Group Relative Policy Optimization (GRPO) using BLEU directly as the reward function. We demonstrate that BLEUBERI-trained models are competitive with models trained via reward model-guided RL across four challenging instruction-following benchmarks and three different base language models. A human evaluation further supports that the quality of BLEUBERI model outputs is on par with those from reward model-aligned models. Moreover, BLEUBERI models generate outputs that are more factually grounded than competing methods. Overall, we show that given access to high-quality reference outputs (easily obtained via existing instruction-following datasets or synthetic data generation), string matching-based metrics are cheap yet effective proxies for reward models during alignment. We release our code and data at https://github.com/lilakk/BLEUBERI.

Boundless Byte Pair Encoding: Breaking the Pre-tokenization Barrier

Pre-tokenization, the initial step in many modern tokenization pipelines, segments text into smaller units called pretokens, typically splitting on whitespace and punctuation. While this process encourages having full, individual words as tokens, it introduces a fundamental limitation in most tokenization algorithms such as Byte Pair Encoding (BPE). Specifically, pre-tokenization causes the distribution of tokens in a corpus to heavily skew towards common, full-length words. This skewed distribution limits the benefits of expanding to larger vocabularies, since the additional tokens appear with progressively lower counts. To overcome this barrier, we propose BoundlessBPE, a modified BPE algorithm that relaxes the pretoken boundary constraint. Our approach selectively merges two complete pretokens into a larger unit we term a superword. Superwords are not necessarily semantically cohesive. For example, the pretokens " of" and " the" might be combined to form the superword " of the". This merging strategy results in a substantially more uniform distribution of tokens across a corpus than standard BPE, and compresses text more effectively, with an approximate 20% increase in bytes per token.

No Free Labels: Limitations of LLM-as-a-Judge Without Human Grounding

LLM-as-a-Judge is a framework that uses an LLM (large language model) to evaluate the quality of natural language text - typically text that is also generated by an LLM. This framework holds great promise due to its relative low-cost, ease of use, and strong correlations with human stylistic preferences. However, LLM Judges have been shown to exhibit biases that can distort their judgments. We evaluate how well LLM Judges can grade whether a given response to a conversational question is correct, an ability crucial to soundly estimating the overall response quality. To do so, we create and publicly release a human-annotated dataset with labels of correctness for 1,200 LLM responses. We source questions from a combination of existing datasets and a novel, challenging benchmark (BFF-Bench) created for this analysis. We demonstrate a strong connection between an LLM's ability to correctly answer a question and grade responses to that question. Although aggregate level statistics might imply a judge has high agreement with human annotators, it will struggle on the subset of questions it could not answer. To address this issue, we recommend a simple solution: provide the judge with a correct, human-written reference answer. We perform an in-depth analysis on how reference quality can affect the performance of an LLM Judge. We show that providing a weaker judge (e.g. Qwen 2.5 7B) with higher quality references reaches better agreement with human annotators than a stronger judge (e.g. GPT-4o) with synthetic references.

How Much is Enough? The Diminishing Returns of Tokenization Training Data

Tokenization, a crucial initial step in natural language processing, is often assumed to benefit from larger training datasets. This paper investigates the impact of tokenizer training data sizes ranging from 1GB to 900GB. Our findings reveal diminishing returns as the data size increases, highlighting a practical limit on how much further scaling the training data can improve tokenization quality. We analyze this phenomenon and attribute the saturation effect to the constraints imposed by the pre-tokenization stage of tokenization. These results offer valuable insights for optimizing the tokenization process and highlight potential avenues for future research in tokenization algorithms.

Are Language Model Logits Calibrated?

Some information is factual (e.g., "Paris is in France"), whereas other information is probabilistic (e.g., "the coin flip will be a [Heads/Tails]."). We believe that good Language Models (LMs) should understand and reflect this nuance. Our work investigates this by testing if LMs' output probabilities are calibrated to their textual contexts. We define model "calibration" as the degree to which the output probabilities of candidate tokens are aligned with the relative likelihood that should be inferred from the given context. For example, if the context concerns two equally likely options (e.g., heads or tails for a fair coin), the output probabilities should reflect this. Likewise, context that concerns non-uniformly likely events (e.g., rolling a six with a die) should also be appropriately captured with proportionate output probabilities. We find that even in simple settings the best LMs (1) are poorly calibrated, and (2) have systematic biases (e.g., preferred colors and sensitivities to word orderings). For example, gpt-4o-mini often picks the first of two options presented in the prompt regardless of the options' implied likelihood, whereas Llama-3.1-8B picks the second. Our other consistent finding is mode-collapse: Instruction-tuned models often over-allocate probability mass on a single option. These systematic biases introduce non-intuitive model behavior, making models harder for users to understand.