activeAgents
MultiChallenge
Metric: % Score (higher is better)
Scale AI benchmark for multi-turn instruction following across instruction retention, inference memory, versioned editing, and self-coherence challenges.
Models ranked
14
tracked on this benchmark
Score band
71.4 – 41.2
best → lowest tracked
Snapshot trend
-16.00
Jun 7 → Oct 8 · 1 models
Leaderboard
Tracked models ranked by % Score (higher is better).
#Model variant and provenanceRelative to leaderScore
1
Gemini 3.1 Pro Preview
71.4Version: MultiChallengeHarness: Not recordedEvaluator: Not recordedObserved: Apr 26, 2026Confidence: Not recordedSource
2
GPT-5.4
69.2Version: MultiChallengeHarness: Not recordedEvaluator: Not recordedObserved: Apr 26, 2026Confidence: Not recordedSource
3
Qwen3.5-397B-A17B
67.6Version: Multi-Challenge leaderboard rank 2 of 28 (accuracy%)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
4
Nemotron 3 Ultra
63.8Version: From official NVIDIA HuggingFace model cardHarness: Not recordedEvaluator: Not recordedObserved: Oct 7, 2026Confidence: Not recordedSource
5
Qwen3.5-122B-A10B
61.5Version: Multi-Challenge leaderboard rank 7 of 28 (accuracy%)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
6
Kimi K2.5
61.4Version: MultiChallengeHarness: Not recordedEvaluator: Not recordedObserved: Apr 26, 2026Confidence: Not recordedSource
7
Gemini 3.1 Flash Lite Preview
60.6Version: MultiChallengeHarness: Not recordedEvaluator: Not recordedObserved: Apr 26, 2026Confidence: Not recordedSource
8
Claude Opus 4.7
58.6Version: MultiChallengeHarness: Not recordedEvaluator: Not recordedObserved: Apr 26, 2026Confidence: Not recordedSource
9
Claude Sonnet 4.6
57.1Version: MultiChallengeHarness: Not recordedEvaluator: Not recordedObserved: Apr 26, 2026Confidence: Not recordedSource
10
MAI-Thinking-1
53.0Version: Multi-Challenge leaderboard rank 15 of 28 (accuracy%)Harness: Not recordedEvaluator: Not recordedObserved: Jun 7, 2026Confidence: Not recordedSource
11
Claude Haiku 4.5
50.5Version: MultiChallengeHarness: Not recordedEvaluator: Not recordedObserved: Apr 26, 2026Confidence: Not recordedSource
12
o4-mini
44.9Version: MultiChallengeHarness: Not recordedEvaluator: Not recordedObserved: Apr 26, 2026Confidence: Not recordedSource
13
MiniMax-M1-80k
44.7Version: MultiChallenge (accuracy)Harness: Not recordedEvaluator: Not recordedObserved: Oct 8, 2026Confidence: Not recordedSource
How to read this benchmark
This benchmark scores models where higher is better. Use scores for directional filtering and shortlisting, not universal quality ranking; then validate pricing, context window, provider availability, and fit for your workload.
Trust this score when
- There is a fresh timestamped snapshot (or multiple snapshots) for this benchmark.
- The model list covers the same version family you can actually deploy today.
- Top candidates overlap with your required routing and feature requirements.
Be cautious when
- There is only one benchmark snapshot or the dataset appears stale.
- The benchmark metric direction is opposite of your decision objective.
- The score difference between options is narrow and likely within implementation variance.
Related benchmarks
Last reviewed: Apr 26, 2026