Remote Labor Index

Real freelance projects completed end to end (Scale AI / CAIS).

Published by

Scale AI / CAISdata via Epoch AI Benchmarking Hub

Category

Agents & tools

Index weight

0.5

Models

12

Data as of

9 Sept 2026

CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.

1. 1Claude Fable 516.1%
2. 2Claude Opus 4.88.3%
3. 3GPT-5.56.3%
4. 4Gemini 3.7 Flash5.0%
5. 5Claude Opus 4.64.2%
6. 6Claude Opus 4.53.8%
7. 7GPT-5.22.5%
8. 8GPT-5.22.1%
9. 9Claude Sonnet 4.52.1%
10. 10GPT-51.7%
11. 11Gemini 3 Pro1.3%
12. 12Gemini 2.5 Pro0.8%

SearchAll providersAnthropicGoogleOpenAIOpen weights12 of 12

#	Model	Score ↓	Run ↕	Index ↕	Published ↕
1	Claude Fable 5Anthropic	16.1%	Fable-5	70.4	–
2	Claude Opus 4.8Anthropic	8.3%	Opus 4.8	64.7	–
3	GPT-5.5OpenAI	6.3%	Codex GPT 5.5	66.6	–
4	Gemini 3.7 FlashGoogle	5.0%	Gemini 3.7 Flash	61.7	–
5	Claude Opus 4.6Anthropic	4.2%	–	63.1	–
6	Claude Opus 4.5Anthropic	3.8%	–	58.0	–
7	GPT-5.2OpenAI	2.5%	–	58.5	–
8	GPT-5.2OpenAI	2.1%	–	61.5	–
9	Claude Sonnet 4.5Anthropic	2.1%	–	52.3	–
10	GPT-5OpenAI	1.7%	–	59.5	–
11	Gemini 3 ProGoogle	1.3%	–	61.6	–
12	Gemini 2.5 ProGoogle	0.8%	–	54.1	–