Remote Labor Index
Real paid freelance projects; share completed to a standard a client would accept.
Agent Index weight
5%
in the Knowledge-work agents pillar
Models tested
18
from the benchmark’s leaderboard and reports
Results
| # | Model | Score | Normalised | Harness | Cost/task | Date |
|---|---|---|---|---|---|---|
| 1 | Claude Fable 5.2 Supported Anthropic | 41.3% | source | — | — | |
| 2 | Claude Opus 5.5 Supported Anthropic | 21.3% | source | — | — | |
| 3 | GPT-6 Astra Supported OpenAI | 20.8% | source | — | — | |
| 4 | Claude Fable 5.1 Supported Anthropic | 17.9% | source | — | — | |
| 5 | Gemini 4 Argon Supported Google | 17.1% | source | — | — | |
| 6 | Claude Fable 5 Supported Anthropic | 16.1% | source | — | — | |
| 7 | DeepSeek V4.1 Flash Estimated DeepSeek | 9.6% | source | — | — | |
| 8 | Claude Opus 4.8 Supported Anthropic | 8.3% | source | — | — | |
| 9 | GPT-5.5 Supported OpenAI | 6.3% | source | — | — | |
| 10 | Gemini 3.8 Flash Supported Google | 5.8% | source | — | — | |
| 11 | Gemini 3.7 Flash Estimated Google | 5.0% | source | — | — | |
| 12 | Claude Opus 4.6 Supported Anthropic | 4.2% | source | — | — | |
| 13 | Claude Opus 4.5 (latest) Supported Anthropic | 3.8% | source | — | — | |
| 14 | GPT-5.2 Supported OpenAI | 2.5% | source | — | — | |
| 15 | Claude Sonnet 4.5 (latest) Supported Anthropic | 2.1% | source | — | — | |
| 16 | GPT-5 Supported OpenAI | 1.7% | source | — | — | |
| 17 | Gemini 3 Pro Preview Supported Google | 1.3% | source | — | — | |
| 18 | Gemini 2.5 Pro Supported Google | 0.8% | source | — | — |
For models without a Remote Labor Index result, the index estimates one from Epoch’s Capabilities Index using a logistic fit on 18 models (residual SD 0.38 on the logit scale). Estimates are labelled and widen the model’s range.