Remote Labor Index
Real freelance projects completed end to end (Scale AI / CAIS).
As of 24 Oct 2026, Claude Opus 5.5 leads Remote Labor Index on BenchLeader with 21.3%, ahead of GPT-6 Astra at 20.8%, across 16 model configurations with a published result.
- Published by
- Scale AI / CAISdata via Epoch AI Benchmarking Hub
- Category
- Agents & tools
- Index weight
- 0.5
- Models
- 16
- Data as of
- 24 Oct 2026
- Newest result
- 22 Sept 2026
CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.
What the test looks like
The Remote Labor Index collects real freelance projects, such as building a website, editing video or preparing a dataset, and asks an agent to deliver the finished work, which professional reviewers judge against the original deliverable.
How it is scored
Percent of projects completed to an acceptable standard, published by Scale AI and the Center for AI Safety.
What to keep in mind
Scores are low for every model and the projects are heterogeneous, so it is better read as "can it do a whole job" than as a fine ranking. Reviews are human and the task set is small.
- 1Claude Opus 5.521.3%
- 2GPT-6 Astra20.8%
- 3Claude Fable 5.117.9%
- 4Claude Fable 515.8%
- 5Claude Opus 4.88.3%
- 6GPT-5.56.3%
- 7Gemini 3.8 Flash5.8%
- 8Gemini 3.7 Flash5.0%
- 9Claude Opus 4.64.2%
- 10Claude Opus 4.53.8%
- 11GPT-5.2 (medium)2.5%
- 12GPT-5.22.1%
- 13Claude Sonnet 4.52.1%
- 14GPT-51.7%
- 15Gemini 3 Pro1.3%
| # | ||||
|---|---|---|---|---|
| 1 |  Claude Opus 5.5 | 21.3% | 62.0 | 2026-09-22 |
| 2 |  GPT-6 Astra | 20.8% | 68.0 | 2026-09-03 |
| 3 | Claude Fable 5.1 | 17.9% | 66.4 | 2026-09-01 |
| 4 | Claude Fable 5 | 15.8% | 67.1 | 2026-06-09 |
| 5 | Claude Opus 4.8 | 8.3% | 60.1 | 2026-05-28 |
| 6 | GPT-5.5 | 6.3% | 60.3 | 2026-04-23 |
| 7 |  Gemini 3.8 Flash | 5.8% | 55.7 | 2026-09-02 |
| 8 | Gemini 3.7 Flash | 5.0% | 53.4 | 2026-08-13 |
| 9 | Claude Opus 4.6 | 4.2% | 62.6 | 2026-02-05 |
| 10 | Claude Opus 4.5 | 3.8% | 57.9 | 2025-11-24 |
| 11 | GPT-5.2medium | 2.5% | 57.6 | 2025-12-11 |
| 12 | GPT-5.2 | 2.1% | 58.0 | 2025-12-11 |
| 13 | Claude Sonnet 4.5 | 2.1% | 53.7 | 2025-09-29 |
| 14 | GPT-5 | 1.7% | 56.2 | 2025-08-07 |
| 15 | Gemini 3 Pro | 1.3% | 60.4 | 2025-11-18 |
| 16 | Gemini 2.5 Pro | 0.8% | 53.9 | 2025-06-05 |
More agents & tools benchmarks · All LLM benchmarks
Cite as: BenchLeader, “Remote Labor Index leaderboard”, data as of 24 Oct 2026, https://www.benchleader.com/benchmarks/rli.
Remote Labor Index: questions
- What does Remote Labor Index measure?
- The Remote Labor Index collects real freelance projects, such as building a website, editing video or preparing a dataset, and asks an agent to deliver the finished work, which professional reviewers judge against the original deliverable. Scores are reported in percent of tasks solved; higher is better.
- Which AI model leads Remote Labor Index?
- Claude Opus 5.5 leads Remote Labor Index with 21.3% as of 24 Oct 2026, ahead of GPT-6 Astra at 20.8%.
- How many models have Remote Labor Index results?
- 16 model configurations have a Remote Labor Index result on BenchLeader, all taken from Scale AI / CAIS via Epoch AI Benchmarking Hub.
- Who runs Remote Labor Index and how often is it updated?
- Remote Labor Index is published by Scale AI / CAIS. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does Remote Labor Index count toward the BenchLeader Index?
- Yes. Remote Labor Index contributes to the agents & tools category of the BenchLeader Index, normalised so that 50 is the average of the evaluated models.