Claude 4.5 Haiku (Reasoning) vs Qwen3.8 27B (Non-reasoning)

Which is better for the money: Claude 4.5 Haiku (Reasoning) by Anthropic or Qwen3.8 27B (Non-reasoning) by Alibaba? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-26.

Trade-off

AnthropicClaude 4.5 Haiku (Reasoning)Off the frontier

Intelligence: 16.90
Total cost: $547

AlibabaQwen3.8 27B (Non-reasoning)Off the frontier

Intelligence: 20.20
Total cost: $2022

Verdict

Depends on your budget

Qwen3.8 27B (Non-reasoning) scores 3.3 points higher (20.2 vs 16.9) but costs 3.7× as much to run the benchmark ($2022 vs $547; flagship vs premium cost classes). They are not alternatives at one budget: pick Qwen3.8 27B (Non-reasoning) when you need intelligence above 16.9, and Claude 4.5 Haiku (Reasoning) when 16.9 is enough.

Metric	Claude 4.5 Haiku (Reasoning)	Qwen3.8 27B (Non-reasoning)	Difference
Intelligence (higher is better)	16.9	20.2	+19.5% Qwen3.8 27B (Non-reas… is 19.5% more intelligent
Total AA cost (lower is better)	$546.96	$2,022.49	+269.8% Qwen3.8 27B (Non-reas… is 3.7× more expensive
End-to-end time (lower is better)	23.69s	14.28s	−39.7% Qwen3.8 27B (Non-reas… is 39.7% faster
Price per 1M tokens (in / out)	$1.00 / $5.00	$0.50 / $3.00
Input modalities	Text, Image	Text, Image, Video

Qwen3.8 27B (Non-reasoning) vs Claude 4.5 Haiku (Reasoning): 19.5% more intelligence, 269.8% more cost to run the AA benchmark, and 39.7% faster end-to-end.

Open this matchup in the interactive picker Claude 4.5 Haiku (Reasoning): what beats it Qwen3.8 27B (Non-reasoning): what beats it

Today's Pareto frontier

Against the frontier

Claude 4.5 Haiku (Reasoning) is off the frontier: at the same or lower cost, GPT-6.1 Sol (High) scores 50.20 for $521.
Qwen3.8 27B (Non-reasoning) is off the frontier: at the same or lower cost, GPT-6.1 Sol (Max) scores 51.80 for $1082.

Claude 4.5 Haiku (Reasoning)Qwen3.8 27B (Non-reasoning)Today's Pareto frontierOpen-weights frontierEvery other model

Matchup

Compare something else

Agents: Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/compare/claude-4-5-haiku-reasoning/qwen3-8-27b-non-reasoning · Markdown: /api/v1/compare/claude-4-5-haiku-reasoning/qwen3-8-27b-non-reasoning?format=md · llms.txt · agent skill