--- title: "evals vs ARES" type: "comparison" canonical_url: "https://www.graphcanon.com/compare/openai-evals-vs-stanford-futuredata-ares" tools: ["openai-evals", "stanford-futuredata-ares"] --- # evals vs ARES *GraphCanon updated Sep 20, 2026* ## Verdict Pick evals if evals is an evaluation framework from OpenAI for assessing large language models and systems built with them. It includes an open-source registry of benchmarks and tools to create custom evaluations; pick ARES if aRES is a Python-based tool for automated evaluation of RAG systems, offering scoring and comparison functionalities. [evals](https://github.com/openai/evals) reports 19k GitHub stars, 3.1k forks, and 333 open issues, last pushed Apr 14, 2026. [ARES](https://ares-ai.vercel.app/) has 733 stars, 67 forks, and 22 open issues, last pushed Mar 28, 2025. Figures are from public GitHub metadata via [evals's repository](https://github.com/openai/evals) and [ARES's repository](https://github.com/stanford-futuredata/ARES). | | [evals](/tools/openai-evals.md) | [ARES](/tools/stanford-futuredata-ares.md) | | --- | --- | --- | | Tagline | Framework for evaluating LLMs and LLM systems with an open-source registry of benchmarks. | Automated Evaluation of RAG Systems | | Stars | 19,398 | 733 | | Forks | 3,077 | 67 | | Open issues | 333 | 22 | | Language | Python | Python | | Adopt for | Evals is an evaluation framework from OpenAI for assessing large language models and systems built with them. It includes an open-source registry of benchmarks and tools to create custom evaluations. | ARES is a Python-based tool for automated evaluation of RAG systems, offering scoring and comparison functionalities. | | Persona | - | - | | Runtime | - | - | | License | Other | Apache-2.0 | | Categories | Evaluation & Observability | Evaluation & Observability | ## Trust and health _Sourced signals - not a safety guarantee. No winner column._ | | [evals](/tools/openai-evals.md) | [ARES](/tools/stanford-futuredata-ares.md) | | --- | --- | --- | | Maintenance | Slowing (36%) | Dormant (18%) | | Days since push | 145d | 538d | | Open issues (now) | 333 | 22 | | Stars delta | +271 (30d) | +2 (30d) | | Open issues delta | +120 (30d) | +1 (30d) | | Full report | [trust report](/tools/openai-evals/trust.md) | [trust report](/tools/stanford-futuredata-ares/trust.md) | ## Shared compatibility - **OpenAI API**: [evals](/tools/openai-evals.md) - OpenAI API; [ARES](/tools/stanford-futuredata-ares.md) - OpenAI API - **Python**: [evals](/tools/openai-evals.md) - Python runtime; [ARES](/tools/stanford-futuredata-ares.md) - Python runtime ## Decision facts: evals - **Adopt for:** Evals is an evaluation framework from OpenAI for assessing large language models and systems built with them. It includes an open-source registry of benchmarks and tools to create custom evaluations. ## Decision facts: ARES - **Requirements:** Requires over 100 GB of available disk space and a compatible GPU like the A100.; Installation of conda and gcc is necessary for setup. - **Adopt for:** ARES is a Python-based tool for automated evaluation of RAG systems, offering scoring and comparison functionalities. ## Choose when ### Choose evals if… - License: evals is Other, ARES is Apache-2.0. - Tags unique to evals: benchmarking, custom eval creation, evaluation-framework, large-language-models. - * When you need a comprehensive set of pre-existing evals and the ability to create your own tailored tests using specific use cases, especially within the OpenAI model ecosystem. ### Choose ARES if… - License: ARES is Apache-2.0, evals is Other. - Requirements: Requires over 100 GB of available disk space and a compatible GPU like the A100.; Installation of conda and gcc is necessary for setup.. - Tags unique to ARES: api integration, automated evaluation, comparison, evaluation criteria. - Use ARES when you need to evaluate RAG systems with specific criteria such as context relevance, answer faithfulness, and answer relevance. ## When NOT to use evals - * When evaluating models or systems that do not benefit from being integrated with the OpenAI API, as some features like direct evals configuration in the OpenAI Dashboard require an OpenAI key. - * If you are looking for an evaluation framework that doesn’t involve external dependencies such as Git Large File Storage (LFS) and specific Python version requirements (Python 3.9 minimum), or if a ## When NOT to use ARES - Do not use ARES if you lack the necessary human-annotated datasets for evaluation criteria or if you do not have a large set of unlabeled triples for scoring. - Avoid ARES if your machine does not meet the requirements, such as having over 100 GB of available disk space and a compatible GPU like the A100. - Do not use ARES if you cannot set up the required environment, including conda and gcc, on your machine. ## Common questions ### What is the difference between evals and ARES? evals: Framework for evaluating LLMs and LLM systems with an open-source registry of benchmarks.. ARES: Automated Evaluation of RAG Systems. See the comparison table for live GitHub stats and shared categories. ### When should I choose evals over ARES? Choose evals over ARES when License: evals is Other, ARES is Apache-2.0; Tags unique to evals: benchmarking, custom eval creation, evaluation-framework, large-language-models; * When you need a comprehensive set of pre-existing evals and the ability to create your own tailored tests using specific use cases, especially within the OpenAI model ecosystem. ### When should I choose ARES over evals? Choose ARES over evals when License: ARES is Apache-2.0, evals is Other; Requirements: Requires over 100 GB of available disk space and a compatible GPU like the A100.; Installation of conda and gcc is necessary for setup.; Tags unique to ARES: api integration, automated evaluation, comparison, evaluation criteria; Use ARES when you need to evaluate RAG systems with specific criteria such as context relevance, answer faithfulness, and answer relevance. ### When should I avoid evals? * When evaluating models or systems that do not benefit from being integrated with the OpenAI API, as some features like direct evals configuration in the OpenAI Dashboard require an OpenAI key. * If you are looking for an evaluation framework that doesn’t involve external dependencies such as Git Large File Storage (LFS) and specific Python version requirements (Python 3.9 minimum), or if a ### When should I avoid ARES? Do not use ARES if you lack the necessary human-annotated datasets for evaluation criteria or if you do not have a large set of unlabeled triples for scoring. Avoid ARES if your machine does not meet the requirements, such as having over 100 GB of available disk space and a compatible GPU like the A100. Do not use ARES if you cannot set up the required environment, including conda and gcc, on your machine. ### Is evals or ARES more popular on GitHub? evals has more GitHub stars (19,398 vs 733). Stars measure visibility, not whether either tool fits your constraints. ### Are evals and ARES open source? Yes - both are open-source projects on GitHub (evals: Other, ARES: Apache-2.0). ### Where can I find alternatives to evals or ARES? GraphCanon lists graph-backed alternatives at [evals alternatives](/tools/openai-evals/alternatives) and [ARES alternatives](/tools/stanford-futuredata-ares/alternatives) ([evals markdown twin](/tools/openai-evals/alternatives.md), [ARES markdown twin](/tools/stanford-futuredata-ares/alternatives.md)), ranked by typed relationship edges rather than popularity votes. ### Is there a machine-readable version of this comparison? Yes. The markdown twin at [this comparison](/compare/openai-evals-vs-stanford-futuredata-ares.md) mirrors this page for agents and LLM crawlers, with the same stats table and FAQ answers. ### Which is better maintained, evals or ARES? evals: Slowing. ARES: Dormant. Compare maintenance labels, days since push, and release cadence in the trust section below - stars alone do not measure maintenance. ### Where are the full trust reports for evals and ARES? GraphCanon publishes per-repo trust reports with dated maintenance, provenance, and scan summaries: [evals trust report](/tools/openai-evals/trust); [ARES trust report](/tools/stanford-futuredata-ares/trust). --- **Machine-readable endpoints** - JSON: [`/api/graphcanon/graph?tool=openai-evals`](/api/graphcanon/graph?tool=openai-evals) - LLM index: [/llms.txt](/llms.txt) - Full corpus: [/llms-full.txt](/llms-full.txt) _GraphCanon - The knowledge graph for AI development. https://www.graphcanon.com/_