HCAST
Agents attempt 189 machine learning, cybersecurity, software engineering and reasoning tasks. 563 human baselines time each task, from one minute to over 8 hours, so success can be read against human task length.
HCAST is the largest task source for the METR time horizon. Time Horizon 1.1 added 73 HCAST tasks. At release agents passed 70 to 80% of tasks under one human hour and under 20% of tasks over four hours.
In the authors’ words
To understand and predict the societal impacts of highly autonomous AI systems, we need benchmarks with grounding, i.e., metrics that directly connect AI performance to real-world effects we care about. We present HCAST (Human-Calibrated Autonomy Software Tasks), a benchmark of 189 machine learning engineering, cybersecurity, software engineering, and general reasoning tasks. We collect 563 human baselines (totaling over 1500 hours) from people skilled in these domains, working under identical conditions as AI agents, which lets us estimate that HCAST tasks take humans between one minute and 8+ hours. Measuring the time tasks take for humans provides an intuitive metric for evaluating AI capabilities, helping answer the question "can an agent be trusted to complete a task that would take a human X hours?" We evaluate the success rates of AI agents built on frontier foundation models, and we find that current agents succeed 70-80% of the time on tasks that take humans less than one hour, and less than 20% of the time on tasks that take humans more than 4 hours.
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Agent task Many steps in a terminal, browser, desktop, app, game or robot.
No scores here yet
We have no scores for HCAST yet.
No lab model card reports it yet
We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.
More in AI research
- MLS-Bench In use Agent must invent one algorithmic change in an ML research scaffold whose gain transfers across settings. Scored relative to human baselines. Claude Fable 5.1 50.3%
- PaperBench In use Agent replicates ICML 2024 papers from scratch. LLM judge grades against author-written hierarchical rubrics.
Qwen 3.8 Max 93% - MLE-bench In use Kaggle-style ML engineering competitions. Agent trains models and is scored against competition leaderboards. GPT-6 Astra 93.8%
- PostTrainBench 1.2 In use CLI agent gets a base model, one H100 and a time budget to post-train it for a target benchmark. Scored on normalized gain. Claude Fable 5.1 44.6%
- KernelBench In use Replace PyTorch operators with faster custom GPU kernels. Scored on correctness and speedup.
MiniMax-M3 28.8% - OpenAI PRs Only in system cards Agents replicate internal OpenAI pull requests from a pre-PR branch and a prompt. Graded by hidden unit tests as pass@1. GPT-5.2 55%