reasoning benchmark
SWE-Lancer
A benchmark for evaluating large language models on real-world freelance software engineering tasks from Upwork. Contains over 1,400 tasks valued at $1 million USD total, ranging from $50 bug fixes to $32,000 feature implementations. Includes both independent engineering tasks graded via end-to-end tests and managerial tasks assessed against original engineering managers' choices.
Updated Aug 7, 2026
Published models4
Registry coverage4
MetricScore
EvidenceB
SWE-Lancer leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | OP | 66.3% | 100.0% | 4 | C | |
| 2 | OP | 37.3% | 66.7% | 4 | C | |
| 3 | OP | 32.6% | 33.3% | 4 | C | |
| 4 | OP | 18.0% | 0.0% | 4 | C |
Score distribution
Top published rows on the benchmark's original scale.
SWE-Lancer highlights
The top published results on this benchmark's own scale.
What is SWE-Lancer?
Definition and scoring fields from the benchmark registry.
A benchmark for evaluating large language models on real-world freelance software engineering tasks from Upwork. Contains over 1,400 tasks valued at $1 million USD total, ranging from $50 bug fixes to $32,000 feature implementations. Includes both independent engineering tasks graded via end-to-end tests and managerial tasks assessed against original engineering managers' choices.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- SWE-Lancer
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- swe-lancer|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about SWE-Lancer.
Which model scores highest on SWE-Lancer?
GPT-5.1 Codex is currently ranked first with 66.3%.
What does SWE-Lancer measure?
A benchmark for evaluating large language models on real-world freelance software engineering tasks from Upwork. Contains over 1,400 tasks valued at $1 million USD total, ranging from $50 bug fixes to $32,000 feature implementations. Includes both independent engineering tasks graded via end-to-end tests and managerial tasks assessed against original engineering managers' choices.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
4 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.