reasoning benchmark
LiveCodeBench v6 Leaderboard
LiveCodeBench v6 is a benchmark for evaluating large language models on code using newly collected programming contest problems.
Updated Sep 5, 2026
Models62
Model coverage62
MetricScore
EvidenceB
LiveCodeBench v6 Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAC | Score91.90% | Percentile100.00% | Participants62 | EvidenceC | Evaluated |
| Rank02 | ModelAC | Score91.90% | Percentile98.36% | Participants62 | EvidenceC | Evaluated |
| Rank03 | ModelAC | Score91.60% | Percentile96.72% | Participants62 | EvidenceC | Evaluated |
| Rank04 | ModelSA | Score90.33% | Percentile95.08% | Participants62 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score90.30% | Percentile93.44% | Participants62 | EvidenceC | Evaluated |
| Rank06 | ModelMA | Score89.60% | Percentile91.80% | Participants62 | EvidenceC | Evaluated |
| Rank07 | ModelAC | Score89.60% | Percentile90.16% | Participants62 | EvidenceC | Evaluated |
| Rank08 | ModelNV | Score89.00% | Percentile88.52% | Participants62 | EvidenceC | Evaluated |
| Rank09 | ModelBY | Score87.80% | Percentile86.89% | Participants62 | EvidenceC | Evaluated |
| Rank10 | ModelMI | Score87.70% | Percentile85.25% | Participants62 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score87.10% | Percentile83.61% | Participants62 | EvidenceC | Evaluated |
| Rank12 | ModelST | Score86.40% | Percentile81.97% | Participants62 | EvidenceC | Evaluated |
| Rank13 | ModelMA | Score85.00% | Percentile80.33% | Participants62 | EvidenceC | Evaluated |
| Rank14 | ModelZA | Score84.90% | Percentile78.69% | Participants62 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score83.90% | Percentile77.05% | Participants62 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score83.60% | Percentile75.41% | Participants62 | EvidenceC | Evaluated |
| Rank17 | ModelMA | Score83.10% | Percentile73.77% | Participants62 | EvidenceC | Evaluated |
| Rank18 | ModelZA | Score82.80% | Percentile72.13% | Participants62 | EvidenceC | Evaluated |
| Rank19 | ModelOP | Score81.90% | Percentile70.49% | Participants62 | EvidenceC | Evaluated |
| Rank20 | ModelBY | Score81.70% | Percentile68.85% | Participants62 | EvidenceC | Evaluated |
| Rank21 | ModelLA | Score81.40% | Percentile67.21% | Participants62 | EvidenceC | Evaluated |
| Rank22 | ModelLA | Score80.70% | Percentile65.57% | Participants62 | EvidenceC | Evaluated |
| Rank23 | ModelAC | Score80.70% | Percentile63.93% | Participants62 | EvidenceC | Evaluated |
| Rank24 | ModelXI | Score80.60% | Percentile62.30% | Participants62 | EvidenceC | Evaluated |
| Rank25 | ModelAC | Score80.40% | Percentile60.66% | Participants62 | EvidenceC | Evaluated |
| Rank26 | ModelGO | Score80.00% | Percentile59.02% | Participants62 | EvidenceC | Evaluated |
| Rank27 | ModelAC | Score78.90% | Percentile57.38% | Participants62 | EvidenceC | Evaluated |
| Rank28 | ModelGO | Score77.10% | Percentile55.74% | Participants62 | EvidenceC | Evaluated |
| Rank29 | ModelIB | Score75.77% | Percentile54.10% | Participants62 | EvidenceC | Evaluated |
| Rank30 | ModelAC | Score74.60% | Percentile52.46% | Participants62 | EvidenceC | Evaluated |
LiveCodeBench v6 Highlights
The leading models and scores on this benchmark.
LiveCodeBench v6 Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for LiveCodeBench v6
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is LiveCodeBench v6?
What LiveCodeBench v6 measures and how its scores work.
LiveCodeBench v6 is a benchmark for large language models for code that collects problems from LeetCode, AtCoder, and CodeForces and annotates them with release dates.
It measures code generation, self-repair, code execution, and test output prediction, reported as a Score ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- LiveCodeBench v6
- Modality
- text
- Primary category
- reasoning
- 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 LiveCodeBench v6.
Which model scores highest on LiveCodeBench v6?
Qwen3.8-Flash-Next is currently ranked first with 91.90%.
What are the top three models on LiveCodeBench v6?
The current leaders are Qwen3.8-Flash-Next (91.90%), Qwen3.8 Flash (91.90%), and Qwen3.7 Max (91.60%).
Which LiveCodeBench v6 model has the lowest official input price?
Step-3.5-Flash has the lowest matched official input price at $0.10 input / $0.30 output per 1M tokens.
Which models are fastest among LiveCodeBench v6 results?
The fastest matched records are GPT OSS 120B High (100.00 tok/s via OpenAI), GLM-4.6 (85.00 tok/s via DeepInfra), and K-EXAONE-236B-A23B (50.00 tok/s via FriendliAI).
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.