Qwen
Efficient

Qwen2.5-3B-Instruct

Micro

3B Qwen2.5, edge-friendly model with surprisingly strong instruction following for its size.

Qwen2.5-3B-Instruct is a efficient AI model from Qwen. It costs $0.0500 per million input tokens and $0.0500 per million output tokens (blended $0.0500/M), with a 32,000-token context window.

RatesUSD per million tokens. Blend is 70% input, 30% output, our calculation.

INPUT
$0.0500/M
per million input tokens
OUTPUT
$0.0500/M
per million output tokens
BLENDED 70/30
$0.0500/M
unchanged since 3 May
CONTEXT
32,000
tokens

Hand-maintained from Qwen's published pricing; no check date on record.

What it is good at
  • Edge / mobile capable
  • Open weights
  • Multilingual at 3B
Typical use cases
  • On-device assistants
  • Edge multilingual chat

Benchmarks

vs. best public score
MATH65.9%
High-school competition math problems. Source: Alibaba (Qwen team) (self-reported), MATH, read 2026-11-10.
HumanEval74.4%
Python function synthesis from docstrings. Source: Alibaba (Qwen team) (self-reported), HumanEval (0-shot), read 2026-11-10.
Hand-curated from each provider's published reports and public leaderboards. Methodology varies across sources, treat as directional rather than authoritative.

Also reported by the provider

Results on benchmarks outside the set we compare across models. They are listed for completeness and are deliberately excluded from the bars above and from every ranking on this site, because a score is only comparable against the same benchmark variant. Why benchmark scores do not compare →

  • MMLU-Redux64.4%self-reported by Alibaba (Qwen team) · 2024-09-19 · Launch blog, instruction-tuned table (the same table gives MMLU-Pro 43.7). Not comparable with the MMLU column, so shown here and never ranked.

More from Qwen

See all 22 →

Frequently asked questions

How much does Qwen2.5-3B-Instruct cost?

Qwen2.5-3B-Instruct costs $0.0500 per million input tokens and $0.0500 per million output tokens, for a blended reference rate of $0.0500 per million tokens.

What is Qwen2.5-3B-Instruct's context window?

Qwen2.5-3B-Instruct supports up to 32,000 tokens of context in a single request.

What is Qwen2.5-3B-Instruct best for?

Qwen2.5-3B-Instruct is well suited to Edge / mobile capable, Open weights and Multilingual at 3B.

Who makes Qwen2.5-3B-Instruct?

Qwen2.5-3B-Instruct is developed and served by Qwen. It was released in Sep 2024.

Terms used on this page