reasoning benchmark
Multi-Challenge Leaderboard
MultiChallenge is a realistic multi-turn conversation evaluation benchmark that challenges frontier LLMs across four key categories: instruction retention (maintaining instructions throughout conversations), inference memory (recalling and connecting details from previous turns), reliable versioned editing (adapting to evolving instructions during collaborative editing), and self-coherence (avoiding contradictions in responses). The benchmark evaluates models on sustained, contextually complex dialogues across diverse topics including travel planning, technical documentation, and professional communication.
Updated Aug 17, 2026
Models29
Model coverage29
MetricScore
EvidenceC
Multi-Challenge Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAM Nova 2 ProAmazon | Score77.7% | Percentile100.0% | Participants29 | EvidenceC | Evaluated |
| Rank02 | ModelAMNova 2 LiteAmazon | Score76.6% | Percentile96.4% | Participants29 | EvidenceC | Evaluated |
| Rank03 | ModelAMNova 2 OmniAmazon | Score75.5% | Percentile92.9% | Participants29 | EvidenceC | Evaluated |
| Rank04 | ModelOP | Score69.6% | Percentile89.3% | Participants29 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score67.6% | Percentile85.7% | Participants29 | EvidenceC | Evaluated |
| Rank06 | ModelNV | Score63.8% | Percentile82.1% | Participants29 | EvidenceC | Evaluated |
| Rank07 | ModelST | Score62.6% | Percentile78.6% | Participants29 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score61.5% | Percentile75.0% | Participants29 | EvidenceC | Evaluated |
| Rank09 | ModelAC | Score60.8% | Percentile71.4% | Participants29 | EvidenceC | Evaluated |
| Rank10 | ModelOP | Score60.4% | Percentile67.9% | Participants29 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score60.0% | Percentile64.3% | Participants29 | EvidenceC | Evaluated |
| Rank12 | ModelNV | Score55.2% | Percentile60.7% | Participants29 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score54.5% | Percentile57.1% | Participants29 | EvidenceC | Evaluated |
| Rank14 | ModelMA | Score54.1% | Percentile53.6% | Participants29 | EvidenceC | Evaluated |
| Rank15 | ModelMA | Score54.1% | Percentile50.0% | Participants29 | EvidenceC | Evaluated |
| Rank16 | ModelMI | Score53.0% | Percentile46.4% | Participants29 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score49.0% | Percentile42.9% | Participants29 | EvidenceC | Evaluated |
| Rank18 | ModelMI | Score44.7% | Percentile39.3% | Participants29 | EvidenceC | Evaluated |
| Rank19 | ModelMI | Score44.7% | Percentile35.7% | Participants29 | EvidenceC | Evaluated |
| Rank20 | ModelOP | Score43.8% | Percentile32.1% | Participants29 | EvidenceC | Evaluated |
| Rank21 | ModelOP | Score43.0% | Percentile28.6% | Participants29 | EvidenceC | Evaluated |
| Rank22 | ModelOP | Score40.3% | Percentile25.0% | Participants29 | EvidenceC | Evaluated |
| Rank23 | ModelOP | Score39.9% | Percentile21.4% | Participants29 | EvidenceC | Evaluated |
| Rank24 | ModelNV | Score38.5% | Percentile17.9% | Participants29 | EvidenceC | Evaluated |
| Rank25 | ModelOP | Score38.3% | Percentile14.3% | Participants29 | EvidenceC | Evaluated |
| Rank26 | ModelOP | Score35.8% | Percentile10.7% | Participants29 | EvidenceC | Evaluated |
| Rank27 | ModelAC | Score33.7% | Percentile7.1% | Participants29 | EvidenceC | Evaluated |
| Rank28 | ModelAC | Score18.9% | Percentile3.6% | Participants29 | EvidenceC | Evaluated |
| Rank29 | ModelOP | Score15.0% | Percentile0.0% | Participants29 | EvidenceC | Evaluated |
Multi-Challenge Highlights
The leading models and scores on this benchmark.
Multi-Challenge Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for Multi-Challenge
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis multi-challenge 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.
- 01AMNova 2 ProAmazon
- Score
- 77.7%
Strengths
- Ranks #1 of 29 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures Multi-Challenge, not total model capability
- 02AMNova 2 LiteAmazon
- Score
- 76.6%
- Price
- $0.33 input / $2.8 output per 1M tokens
Strengths
- Ranks #2 of 29 compared models
- 96th percentile on this benchmark
- C evidence result
Considerations
- This result measures Multi-Challenge, not total model capability
- 03AMNova 2 OmniAmazon
- Score
- 75.5%
Strengths
- Ranks #3 of 29 compared models
- 93th percentile on this benchmark
- C evidence result
Considerations
- This result measures Multi-Challenge, not total model capability
- 04OPGPT-5OpenAI
- Score
- 69.6%
- Price
- $1.3 input / $10 output per 1M tokens
- Speed
- Up to 100 tok/s via OpenAI
Strengths
- Ranks #4 of 29 compared models
- 89th percentile on this benchmark
- C evidence result
Considerations
- This result measures Multi-Challenge, not total model capability
- 05ACQwen3.5-397B-A17BAlibaba Cloud / Qwen Team
- Score
- 67.6%
- Price
- $0.60 input / $3.6 output per 1M tokens
Strengths
- Ranks #5 of 29 compared models
- 86th percentile on this benchmark
- C evidence result
Considerations
- This result measures Multi-Challenge, not total model capability
Selection summary
Best AI Models for Multi-Challenge
Nova 2 Pro currently leads Multi-Challenge with 77.7%. 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 Multi-Challenge?
What Multi-Challenge measures and how its scores work.
MultiChallenge is a realistic multi-turn conversation evaluation benchmark that challenges frontier LLMs across four key categories: instruction retention (maintaining instructions throughout conversations), inference memory (recalling and connecting details from previous turns), reliable versioned editing (adapting to evolving instructions during collaborative editing), and self-coherence (avoiding contradictions in responses). The benchmark evaluates models on sustained, contextually complex dialogues across diverse topics including travel planning, technical documentation, and professional communication.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of C.
- Family
- Multi-Challenge
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- multichallenge|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about Multi-Challenge.
Which model scores highest on Multi-Challenge?
Nova 2 Pro is currently ranked first with 77.7%.
What does Multi-Challenge measure?
MultiChallenge is a realistic multi-turn conversation evaluation benchmark that challenges frontier LLMs across four key categories: instruction retention (maintaining instructions throughout conversations), inference memory (recalling and connecting details from previous turns), reliable versioned editing (adapting to evolving instructions during collaborative editing), and self-coherence (avoiding contradictions in responses). The benchmark evaluates models on sustained, contextually complex dialogues across diverse topics including travel planning, technical documentation, and professional communication.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
29 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.