Inkling Small (xHigh): benchmark results

Provider: Thinking Machines. Released 2026-07-30. Access: Open.

Aggregate Rating 1654 ± 33, rank #532 of 2245 rated models, from 16 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Epoch AI - GPQA Diamond88.51Accuracy (%)79.1
SecureBio BioTIER Permit99.8Benign prompts answered (%)78.9
OTIS Mock AIME 2024-2590Accuracy (%)72.5
SecureBio HPCT54.91Accuracy (%)63.4
ARC-AGI-184Accuracy (%)58.1
ARC-AGI-240.14Accuracy (%)56.5
FrontierMath - Tiers 1-3 (v2)46.32Accuracy (%, 285 private v2 problems)56.1
SecureBio ABLE Full57.5Accuracy (%)52.6
SecureBio ABLE Tasks With Baselines Only34.49Accuracy (%)52.6
FrontierMath - Tier 4 (v2)17.07Accuracy (%, 41 private v2 problems)28.4
SecureBio BioTIER Refuse24.9Hazardous prompts refused (%)27.6
SecureBio VCT v234.46Accuracy (%)15.4

Interactive version: theaggregate.ai/model?slug=inkling-small-xhigh · How It Works · Data refreshed daily, snapshot 2026-11-16.