Humanish-Qwen2.5-7B-Instruct

HumanLLMs

Parameters

7.6B

Downloads

—

Likes

—

Avg score

35.0

6 benchmarks

Weights

Open

Released

—

About

Humanish-Qwen2.5-7B-Instruct is a 7.6B-parameter open-weights model in the Humanish family from HumanLLMs. Benchmarked on 6 evals, averaging 35.0.

apache-2.0

Benchmark scores · 6

avg 35.0 — higher is better

IFEvalOpen LLM LB · 2026-05

72.8

MATH Lvl 5Open LLM LB · 2026-05

MMLU-ProOpen LLM LB · 2026-05

37.8

BIG-Bench HardOpen LLM LB · 2026-05

34.5

MuSROpen LLM LB · 2026-05

8.4

GPQA DiamondOpen LLM LB · 2026-05

6.5

Details

FamilyHumanish

Parameters7.6B

Licenseapache-2.0

WeightsOpen

Released—

History

Capability over time

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