reasoning benchmark
SWE-bench Multilingual Leaderboard
A multilingual benchmark for issue resolving in software engineering that covers Java, TypeScript, JavaScript, Go, Rust, C, and C++. Contains 1,632 high-quality instances carefully annotated from 2,456 candidates by 68 expert annotators, designed to evaluate Large Language Models across diverse software ecosystems beyond Python.
Updated Aug 17, 2026
Models38
Model coverage38
MetricScore
EvidenceB
SWE-bench Multilingual Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAN | Score87.3% | Percentile100.0% | Participants38 | EvidenceC | Evaluated |
| Rank02 | ModelAN | Score84.4% | Percentile97.3% | Participants38 | EvidenceC | Evaluated |
| Rank03 | ModelPO | Score78.5% | Percentile94.6% | Participants38 | EvidenceC | Evaluated |
| Rank04 | ModelAN | Score78.3% | Percentile91.9% | Participants38 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score78.3% | Percentile89.2% | Participants38 | EvidenceC | Evaluated |
| Rank06 | ModelAN | Score77.8% | Percentile86.5% | Participants38 | EvidenceC | Evaluated |
| Rank07 | ModelMA | Score76.7% | Percentile83.8% | Participants38 | EvidenceC | Evaluated |
| Rank08 | ModelMI | Score76.5% | Percentile81.1% | Participants38 | EvidenceC | Evaluated |
| Rank09 | ModelDE | Score76.2% | Percentile78.4% | Participants38 | EvidenceC | Evaluated |
| Rank10 | ModelTE | Score75.8% | Percentile75.7% | Participants38 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score75.8% | Percentile73.0% | Participants38 | EvidenceC | Evaluated |
| Rank12 | ModelAC | Score73.8% | Percentile70.3% | Participants38 | EvidenceC | Evaluated |
| Rank13 | ModelDE | Score73.3% | Percentile67.6% | Participants38 | EvidenceC | Evaluated |
| Rank14 | ModelMA | Score73.0% | Percentile64.9% | Participants38 | EvidenceC | Evaluated |
| Rank15 | ModelMI | Score72.5% | Percentile62.2% | Participants38 | EvidenceC | Evaluated |
| Rank16 | ModelXI | Score71.7% | Percentile59.5% | Participants38 | EvidenceC | Evaluated |
| Rank17 | ModelXI | Score71.7% | Percentile56.8% | Participants38 | EvidenceC | Evaluated |
| Rank18 | ModelAC | Score71.3% | Percentile54.0% | Participants38 | EvidenceC | Evaluated |
| Rank19 | ModelDE | Score70.2% | Percentile51.4% | Participants38 | EvidenceC | Evaluated |
| Rank20 | ModelDE | Score70.2% | Percentile48.6% | Participants38 | EvidenceC | Evaluated |
| Rank21 | ModelDE | Score70.2% | Percentile46.0% | Participants38 | EvidenceC | Evaluated |
| Rank22 | ModelAC | Score69.3% | Percentile43.2% | Participants38 | EvidenceC | Evaluated |
| Rank23 | ModelNV | Score67.7% | Percentile40.5% | Participants38 | EvidenceC | Evaluated |
| Rank24 | ModelAC | Score67.2% | Percentile37.8% | Participants38 | EvidenceC | Evaluated |
| Rank25 | ModelZA | Score66.7% | Percentile35.1% | Participants38 | EvidenceC | Evaluated |
| Rank26 | ModelMI | Score65.5% | Percentile32.4% | Participants38 | EvidenceC | Evaluated |
| Rank27 | ModelPO | Score63.1% | Percentile29.7% | Participants38 | EvidenceC | Evaluated |
| Rank28 | ModelMA | Score61.1% | Percentile27.0% | Participants38 | EvidenceC | Evaluated |
| Rank29 | ModelDE | Score57.9% | Percentile24.3% | Participants38 | EvidenceC | Evaluated |
| Rank30 | ModelMI | Score56.5% | Percentile21.6% | Participants38 | EvidenceC | Evaluated |
| Rank31 | ModelAC | Score54.7% | Percentile18.9% | Participants38 | EvidenceC | Evaluated |
| Rank32 | ModelDE | Score54.5% | Percentile16.2% | Participants38 | EvidenceC | Evaluated |
| Rank33 | ModelMA | Score47.3% | Percentile13.5% | Participants38 | EvidenceC | Evaluated |
| Rank34 | ModelMA | Score47.3% | Percentile10.8% | Participants38 | EvidenceC | Evaluated |
| Rank35 | ModelNV | Score45.8% | Percentile8.1% | Participants38 | EvidenceC | Evaluated |
| Rank36 | ModelNV | Score39.3% | Percentile5.4% | Participants38 | EvidenceC | Evaluated |
| Rank37 | ModelME | Score38.1% | Percentile2.7% | Participants38 | EvidenceC | Evaluated |
| Rank38 | ModelDE | Score30.5% | Percentile0.0% | Participants38 | EvidenceC | Evaluated |
SWE-bench Multilingual Highlights
The leading models and scores on this benchmark.
SWE-bench Multilingual Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for SWE-bench Multilingual
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis swe-bench multilingual AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01ANClaude Mythos PreviewAnthropic
- Score
- 87.3%
Strengths
- Ranks #1 of 38 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures SWE-bench Multilingual, not total model capability
- 02ANClaude Opus 4.8Anthropic
- Score
- 84.4%
- Price
- $5.0 input / $25 output per 1M tokens
- Speed
- Up to 46 tok/s via Anthropic
Strengths
- Ranks #2 of 38 compared models
- 97th percentile on this benchmark
- C evidence result
Considerations
- This result measures SWE-bench Multilingual, not total model capability
- 03POLaguna S 2.1Poolside
- Score
- 78.5%
Strengths
- Ranks #3 of 38 compared models
- 95th percentile on this benchmark
- C evidence result
Considerations
- This result measures SWE-bench Multilingual, not total model capability
- 04ANClaude Sonnet 5Anthropic
- Score
- 78.3%
- Price
- $2.0 input / $10 output per 1M tokens
- Speed
- Up to 11 tok/s via Anthropic
Strengths
- Ranks #4 of 38 compared models
- 92th percentile on this benchmark
- C evidence result
Considerations
- This result measures SWE-bench Multilingual, not total model capability
- 05ACQwen3.7 MaxAlibaba Cloud / Qwen Team
- Score
- 78.3%
- Price
- $2.5 input / $7.5 output per 1M tokens
- Speed
- Up to 5.8 tok/s via Together
Strengths
- Ranks #5 of 38 compared models
- 89th percentile on this benchmark
- C evidence result
Considerations
- This result measures SWE-bench Multilingual, not total model capability
Selection summary
Best AI Models for SWE-bench Multilingual
Claude Mythos Preview currently leads SWE-bench Multilingual with 87.3%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.
What is SWE-bench Multilingual?
What SWE-bench Multilingual measures and how its scores work.
A multilingual benchmark for issue resolving in software engineering that covers Java, TypeScript, JavaScript, Go, Rust, C, and C++. Contains 1,632 high-quality instances carefully annotated from 2,456 candidates by 68 expert annotators, designed to evaluate Large Language Models across diverse software ecosystems beyond Python.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- SWE-bench Multilingual
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- swe-bench-multilingual|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about SWE-bench Multilingual.
Which model scores highest on SWE-bench Multilingual?
Claude Mythos Preview is currently ranked first with 87.3%.
What does SWE-bench Multilingual measure?
A multilingual benchmark for issue resolving in software engineering that covers Java, TypeScript, JavaScript, Go, Rust, C, and C++. Contains 1,632 high-quality instances carefully annotated from 2,456 candidates by 68 expert annotators, designed to evaluate Large Language Models across diverse software ecosystems beyond Python.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
38 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.