Qwen3 0.6B (Non-reasoning)

Qwen3 0.6B (Non-reasoning)

Alibaba · Released Apr 28, 2025 Professional
Intelligence #14 / 590
82.0 Our Score
Speed #25 / 279
222.1 tokens / sec
Input #224 / 592
$0.110 per 1M tokens
Output #250 / 592
$0.420 per 1M tokens
Context
— Not reported

Analysis Summary

Qwen3 0.6B Non-reasoning is Alibaba's smallest model in the Qwen3 family, with an intelligence index of 1.0, GPQA of 0.231, and MMLU-Pro of 0.231. At $0.11 input and $0.42 output per million tokens, it is among the cheapest models available, but capability is correspondingly limited.

For businesses, it is only viable for the simplest, most narrowly defined tasks: keyword classification, basic intent detection, or lightweight text formatting where a larger model would be wasteful. Coding capability is very limited at LiveCodeBench 0.073, and reasoning depth is insufficient for any complex workflow. The agentic index of 14.6 is surprisingly moderate but unlikely to translate to reliable real-world agent performance at this intelligence level.

A -4 point regional penalty applies. Teams with high-volume, low-complexity automation needs and tight cost constraints may find a use case, but for most business applications a slightly larger model will deliver far better results at a still-minimal cost.

Assessed June 30, 2026

Editorial notes

Qwen3 0.6B Non-reasoning from Alibaba is an ultra-compact model priced at $0.11 input, suited only to the simplest classification or extraction tasks; reasoning and coding capability are very limited at this scale.

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?

Performance Profile

Intelligence0.7Technical1.6Value7Content2.4
Intelligence 0.7/10
Technical 1.6/10
Content 2.4/10
Value 7/10

How Qwen3 0.6B (Non-reasoning) compares

Qwen3 0.6B (Non-reasoning) ranks #380 of 385 AI models we track for overall intelligence, #226 of 293 for agentic tasks. At $0.11 per million input tokens it is cheaper than 62% of comparable models.

Performance Indices

Source: Artificial Analysis

1 Intelligence Index
14.6 Agentic Index
10.3 Math Index

Benchmark Scores

Intelligence

GPQA Diamond 23.1% Graduate-level scientific reasoning
HLE 5.2% Humanity's Last Exam
MMLU Pro 23.1% Multi-task language understanding
MATH 500 52.1% Mathematical problem-solving
AIME 1.7% Competition mathematics
AIME 2025 10.3% Competition mathematics (2025)
SciCode 4.1% Scientific computing

Technical

LiveCodeBench 7.3% Live coding evaluation
τ²-Bench 14.6% Conversational agent benchmark

Content

IFBench 21.9% Instruction following

Benchmark data from Artificial Analysis and Hugging Face

How does Qwen3 0.6B (Non-reasoning) stack up?

Compare side-by-side with other professional models.

Compare Models

Model Information

ProviderAlibaba
Release Date April 28, 2025
Status Active

Pricing

Token Type Cost per 1M tokens Cost per 1K tokens
Input $0.11 $0.000110
Output $0.42 $0.000420

Frequently asked questions about Qwen3 0.6B (Non-reasoning)

How much does Qwen3 0.6B (Non-reasoning) cost?

Qwen3 0.6B (Non-reasoning) costs $0.11 per million input tokens and $0.42 per million output tokens.

Who created Qwen3 0.6B (Non-reasoning)?

Qwen3 0.6B (Non-reasoning) is developed by Alibaba and was released on April 28, 2025.