agents benchmark
FrontierSWE Leaderboard
FrontierSWE evaluates whether an agent can complete open-ended technical projects spanning systems optimization, large-scale code construction, and applied ML research over hours to tens of hours.
Updated Sep 7, 2026
Models16
Model coverage16
MetricScore
EvidenceB
FrontierSWE Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAN | Score90.00% | Percentile100.00% | Participants16 | EvidenceB | Evaluated |
| Rank02 | ModelMA | Score81.20% | Percentile93.33% | Participants16 | EvidenceC | Evaluated |
| Rank03 | ModelZA | Score78.10% | Percentile86.67% | Participants16 | EvidenceC | Evaluated |
| Rank04 | ModelAN | Score75.00% | Percentile80.00% | Participants16 | EvidenceB | Evaluated |
| Rank05 | ModelZA | Score74.00% | Percentile73.33% | Participants16 | EvidenceB | Evaluated |
| Rank06 | ModelAC | Score73.50% | Percentile66.67% | Participants16 | EvidenceC | Evaluated |
| Rank07 | ModelOP | Score73.00% | Percentile60.00% | Participants16 | EvidenceB | Evaluated |
| Rank08 | ModelAN | Score63.00% | Percentile53.33% | Participants16 | EvidenceB | Evaluated |
| Rank09 | ModelAN | Score56.00% | Percentile46.67% | Participants16 | EvidenceB | Evaluated |
| Rank10 | ModelOP | Score54.00% | Percentile40.00% | Participants16 | EvidenceB | Evaluated |
| Rank11 | ModelGO | Score40.00% | Percentile33.33% | Participants16 | EvidenceB | Evaluated |
| Rank12 | ModelZA | Score31.00% | Percentile26.67% | Participants16 | EvidenceB | Evaluated |
| Rank13 | ModelDE | Score29.00% | Percentile20.00% | Participants16 | EvidenceB | Evaluated |
| Rank14 | ModelMA | Score27.00% | Percentile13.33% | Participants16 | EvidenceB | Evaluated |
| Rank15 | ModelMA | Score26.00% | Percentile6.67% | Participants16 | EvidenceB | Evaluated |
| Rank16 | ModelAC | Score22.00% | Percentile0.00% | Participants16 | EvidenceB | Evaluated |
FrontierSWE Highlights
The leading models and scores on this benchmark.
FrontierSWE Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for FrontierSWE
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is FrontierSWE?
What FrontierSWE measures and how its scores work.
FrontierSWE is an agents benchmark for open-ended technical projects at the scale of hours to tens of hours.
It measures agent performance on systems optimization, large-scale code construction, and applied ML research, reported as a dominance score where higher is better.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- FrontierSWE
- Modality
- text
- Primary category
- agents
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about FrontierSWE.
Which model scores highest on FrontierSWE?
Claude Fable 5 is currently ranked first with 90.00%.
What are the top three models on FrontierSWE?
The current leaders are Claude Fable 5 (90.00%), Kimi K3 (81.20%), and GLM-5.3 (78.10%).
Which FrontierSWE model has the lowest official input price?
Qwen3.6 Plus has the lowest matched official input price at $0.50 input / $3.0 output per 1M tokens.
Which models are fastest among FrontierSWE results?
The fastest matched records are GLM-5.3 (472.69 tok/s via FriendliAI), GPT-5.5 (134.94 tok/s via OpenAI), and Claude Opus 4.6 (123.15 tok/s via Anthropic).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.