DeepHermes 3 - Mistral 24B Preview

Nous Research

Verified against Artificial Analysis · Jul 13, 2026

Capability

percentile index

Price /M

$0 / $0

in / out

Context

—

Avg score

33.6

6 benchmarks

AA Intelligence

5.3

index

Weights

Open

Released

Mar 13, 2025

About

DeepHermes 3 - Mistral 24B Preview is a open-weights model in the DeepHermes family from Nous Research. Benchmarked on 6 evals, averaging 33.6.

Benchmark scores · 6

avg 33.6 — higher is better

MATHartificial-analysis · 2026-07

59.5

MMLU-Proartificial-analysis · 2026-07

GPQA Diamondartificial-analysis · 2026-07

38.2

SciCodeartificial-analysis · 2026-07

22.8

LiveCodeBenchartificial-analysis · 2026-07

19.5

Humanity's Last Examartificial-analysis · 2026-07

3.9

Details

FamilyDeepHermes

Parameters—

License—

WeightsOpen

ReleasedMar 13, 2025

History

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers

Artificial Analysis$0/M