Refresh Models —Deals —Regulatory —Sources —

Claude 3.5 Haiku vs HyperNova 60B 2605 (high, based on gpt-oss-120b)

17 SHARED BENCHMARKS

Across 17 shared benchmarks, Claude 3.5 Haiku scores higher on 3 and HyperNova 60B 2605 (high, based on gpt-oss-120b) on 14. The widest gap is HLE, where HyperNova 60B 2605 (high, based on gpt-oss-120b) scores 16.6 against 3.6.

ANTHROPICVSOPENAI17 SHARED3–14 HEAD-TO-HEAD

BenchmarkClaude 3.5 HaikuMARGINHyperNova 60B 2605 (high, based on gpt-oss-120b)

AA Agentic Index1.96.8

AA Intelligence12.318.3

AA-LCR25.335.3

AA-Omniscience-22.5-54.6

Artificial Analysis Coding Index15.923.2

GPQA Diamond41.673.3

HLE3.616.6

IFBench42.866.5

OmniScience Accuracy13.215.2

OmniScience Non-Hallucination58.917.7

TauBench V3 - Banking5.85.6

Terminal-Bench 2.110.118.4

Terminal-Bench Hard2.323.5

τ²-Bench Telecom (AA run)24.663.2

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.