HumanEvalFIM-Average

Average evaluation of HumanEval Fill-in-the-Middle benchmark variants (single-line, multi-line, random-span) for assessing code infilling capabilities of language models

Codestral-22B from Mistral AI currently leads the HumanEvalFIM-Average leaderboard with a score of 0.916 across 1 evaluated AI models.

Paper
About this benchmark

What HumanEvalFIM-Average measures

HumanEvalFIM-Average is a text benchmark that evaluates large language models on general tasks. LLM Stats tracks 1 model on this benchmark, with a maximum possible score of 1. Current average across reported models is 0.9, with the leader reaching 0.9.

Compare leaders on the best AI for general leaderboards.

Publication

Paper
Efficient Training of Language Models to Fill in the Middle
Authors
Mohammad Bavarian, Heewoo Jun, Nikolas Tezak, John Schulman, and 3 others
Published

Abstract

We show that autoregressive language models can learn to infill text after we apply a straightforward transformation to the dataset, which simply moves a span of text from the middle of a document to its end. While this data augmentation has garnered much interest in recent years, we provide extensive evidence that training models with a large fraction of data transformed in this way does not harm the original left-to-right generative capability, as measured by perplexity and sampling evaluations across a wide range of scales. Given the usefulness, simplicity, and efficiency of training models to fill-in-the-middle (FIM), we suggest that future autoregressive language models be trained with FIM by default. To this end, we run a series of ablations on key hyperparameters, such as the data transformation frequency, the structure of the transformation, and the method of selecting the infill span. We use these ablations to prescribe strong default settings and best practices to train FIM models. We have released our best infilling model trained with best practices in our API, and release our infilling benchmarks to aid future research.

Mistral AICodestral-22B leads with 91.6%.

Progress Over Time

Interactive timeline showing model performance evolution on HumanEvalFIM-Average

State-of-the-art frontier
Open
Proprietary

HumanEvalFIM-Average Leaderboard

1 models
ContextCostLicense
1
Mistral AI
Mistral AI
22B
Notice missing or incorrect data?

FAQ

Common questions about HumanEvalFIM-Average.

What is the HumanEvalFIM-Average benchmark?

Average evaluation of HumanEval Fill-in-the-Middle benchmark variants (single-line, multi-line, random-span) for assessing code infilling capabilities of language models

What is the HumanEvalFIM-Average leaderboard?

The HumanEvalFIM-Average leaderboard ranks 1 AI models based on their performance on this benchmark. Currently, Codestral-22B by Mistral AI leads with a score of 0.916. The average score across all models is 0.916.

What is the highest HumanEvalFIM-Average score?

The highest HumanEvalFIM-Average score is 0.916, achieved by Codestral-22B from Mistral AI.

How many models are evaluated on HumanEvalFIM-Average?

1 models have been evaluated on the HumanEvalFIM-Average benchmark, with 0 verified results and 1 self-reported results.

Where can I find the HumanEvalFIM-Average paper?

The HumanEvalFIM-Average paper is available at https://arxiv.org/abs/2207.14255. The paper details the methodology, dataset construction, and evaluation criteria.

What categories does HumanEvalFIM-Average cover?

HumanEvalFIM-Average is categorized under general. The benchmark evaluates text models.

What is the best open-source model on HumanEvalFIM-Average?

Codestral-22B by Mistral AI is the top-ranked open-source model on HumanEvalFIM-Average, with a score of 0.916 (rank #1).

How recent are the HumanEvalFIM-Average leaderboard results?

The HumanEvalFIM-Average leaderboard was last updated in June 2026 and currently includes 1 evaluated models.

More evaluations to explore

Related benchmarks in the same category

View all general
GPQA

A challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry. Questions are Google-proof and extremely difficult, with PhD experts reaching 65% accuracy.

general
224 models
MMLU-Pro

A more robust and challenging multi-task language understanding benchmark that extends MMLU by expanding multiple-choice options from 4 to 10, eliminating trivial questions, and focusing on reasoning-intensive tasks. Features over 12,000 curated questions across 14 domains and causes a 16-33% accuracy drop compared to original MMLU.

general
127 models
MMLU

Massive Multitask Language Understanding benchmark testing knowledge across 57 diverse subjects including STEM, humanities, social sciences, and professional domains

general
100 models
LiveCodeBench

LiveCodeBench is a holistic and contamination-free evaluation benchmark for large language models for code. It continuously collects new problems from programming contests (LeetCode, AtCoder, CodeForces) and evaluates four different scenarios: code generation, self-repair, code execution, and test output prediction. Problems are annotated with release dates to enable evaluation on unseen problems released after a model's training cutoff.

general
74 models
IFEval

Instruction-Following Evaluation (IFEval) benchmark for large language models, focusing on verifiable instructions with 25 types of instructions and around 500 prompts containing one or more verifiable constraints

general
64 models
MMMU

MMMU (Massive Multi-discipline Multimodal Understanding) is a benchmark designed to evaluate multimodal models on college-level subject knowledge and deliberate reasoning. Contains 11.5K meticulously collected multimodal questions from college exams, quizzes, and textbooks, covering six core disciplines: Art & Design, Business, Science, Health & Medicine, Humanities & Social Science, and Tech & Engineering across 30 subjects and 183 subfields.

generalmultimodal
62 models