Pairwise benchmark snapshot · Aug 17, 2026noindex (thin pair)

Qwen2.5 Coder 32B Instruct vs Qwen: Qwen3.5-27B benchmark

Name: Qwen2.5 Coder 32B Instruct vs Qwen: Qwen3.5-27B AI benchmark comparison
Creator: CompareLLM
License: https://creativecommons.org/licenses/by/4.0/

In this head-to-head showdown, Qwen2.5 Coder 32B Instruct is more budget-friendly at $1/1M per 1M output tokens. Review the complete breakdown below to determine which model best fits your performance and budget requirements.

Share Analysis:Share:

WhatsApp Telegram X LinkedIn

Executive Verdict Summary

Qwen: Qwen3.5-27B holds empirical advantage (2 of 3 metrics)

Coding / SWESWE-bench

Qwen2.5 Coder 32B InstructTop code solve rate

ReasoningGPQA / Live

Qwen2.5 Coder 32B InstructTop logical accuracy

Best Value$/1M Tok

Qwen2.5 Coder 32B InstructLowest API billing cost

Speed / TokTok/Sec

Qwen2.5 Coder 32B InstructFastest stream rate

Verified Advantage Breakdown

3 benchmarks evaluated across capability, speed, and cost

Qwen2.5 Coder 32B Instruct (1)Qwen: Qwen3.5-27B (2)

Qwen2.5 Coder 32B Instruct

1 of 3 Wins

✓Output price

Qwen: Qwen3.5-27B

2 of 3 Wins

✓Input price✓Context window

Alibaba

Leads 1 of 3

Qwen2.5 Coder 32B Instruct

Auto-discovered from OpenRouter (qwen/qwen-2.5-coder-32b-instruct). Preview until a second source matches.

Elo 1,425In: $0.66 · Out: $1/1M

Alibaba

Leads 2 of 3

Qwen: Qwen3.5-27B

Auto-discovered from OpenRouter (qwen/qwen3.5-27b). Preview until a second source matches.

In: $0.195 · Out: $1.56/1M

Top Rival Showdowns for Qwen2.5 Coder 32B Instruct

All Matchups

Compare vs:vs Claude Opus 4.5 vs GPT-5 vs Gemini 3 Pro vs Grok 4.6 vs DeepSeek V4 Pro

Preference Leader

No shared data

Single model data

Throughput Leader

No shared data

No latency data

Value per Dollar LeaderQwen2.5 Coder 32B Instruct Wins

Qwen2.5 Coder 32B Instruct

$1 / 1M output

Capability Percentiles

Relative percentile scores computed across all active models in the benchmark catalog.

Multi-Dimensional Capability Radar

Leaders Matchup Capability Radar

Comparing top standard models across 6 key capability dimensions. Tap any spoke or dot to inspect.

Tap any node to inspect

0–100 %ile

Overall Matchup Breakdown

Category wins across reasoning intelligence, generation speed, and token cost.

Qwen2.5 Coder 32B Instruct (5)vsQwen: Qwen3.5-27B (2)

Qwen2.5 Coder 32B Instruct: 5W (71%)Qwen: Qwen3.5-27B: 2W (29%)

← Qwen2.5 Coder 32B InstructQwen: Qwen3.5-27B →

🏆Qwen2.5 Coder 32B Instruct Leads(4/5)

Intelligence & Reasoning

Preference Elo, Coding proficiency, SWE-bench & LiveBench accuracy

Benchmark

Qwen2.5 Coder 32B InstructvsQwen: Qwen3.5-27B

Preference Elo

1,425vs—

Coding Elo

1,495vs—

SWE-bench

65.2%vs—

LiveBench

64.8%vs—

GPQA Diamond

—vs—

Qwen2.5 Coder 32B Instruct: 4WQwen: Qwen3.5-27B: 0W

Tied (0-0)

Speed & Latency

Generation throughput and time to first token responsiveness

Benchmark

Qwen2.5 Coder 32B InstructvsQwen: Qwen3.5-27B

Output speed

—vs—

Time to first token

—vs—

Qwen2.5 Coder 32B Instruct: 0WQwen: Qwen3.5-27B: 0W

🏆Qwen: Qwen3.5-27B Leads(2/3)

Pricing & Capacity

Cost per million tokens and max context window length

Benchmark

Qwen2.5 Coder 32B InstructvsQwen: Qwen3.5-27B

Output price

$1/1Mvs$1.56/1M+$0.56/1M

Input price

$0.66/1Mvs$0.195/1M+$0.465/1M

Context window

33kvs262k+229k

Qwen2.5 Coder 32B Instruct: 1WQwen: Qwen3.5-27B: 2W

Side-by-Side Benchmark Matrix

Qwen2.5 Coder 32B InstructQwen: Qwen3.5-27B

Preference Elo

Qwen2.5 Coder 32B Instruct1,425

Qwen: Qwen3.5-27B—

Coding Elo

Qwen2.5 Coder 32B Instruct1,495

Qwen: Qwen3.5-27B—

LiveBench

Qwen2.5 Coder 32B Instruct64.8%

Qwen: Qwen3.5-27B—

SWE-bench

Qwen2.5 Coder 32B Instruct65.2%

Qwen: Qwen3.5-27B—

Input priceQwen: Qwen3.5-27B +$0.465/1M

Qwen2.5 Coder 32B Instruct$0.66/1M

Qwen: Qwen3.5-27B$0.195/1M

Output priceQwen2.5 Coder 32B Instruct +$0.56/1M

Qwen2.5 Coder 32B Instruct$1/1M

Qwen: Qwen3.5-27B$1.56/1M

Context windowQwen: Qwen3.5-27B +229k

Qwen2.5 Coder 32B Instruct33k

Qwen: Qwen3.5-27B262k

Benchmark	Qwen2.5 Coder 32B Instruct	Qwen: Qwen3.5-27B	Advantage Delta
Preference Elo	1,425 lmarena · Aug 17, 2026	—	—
Coding Elo	1,495 lmarena · Aug 17, 2026	—	—
LiveBench	64.8% livebench · Aug 17, 2026	—	—
SWE-bench	65.2% swebench · Aug 17, 2026	—	—
Input price	$0.66/1M openrouter · Aug 17, 2026	$0.195/1M openrouter · Aug 17, 2026	+$0.465/1M
Output price	$1/1M openrouter · Aug 17, 2026	$1.56/1M openrouter · Aug 17, 2026	+$0.56/1M
Context window	33k openrouter · Aug 17, 2026	262k openrouter · Aug 17, 2026	+229k

Workload Cost & Savings Calculator

Simulate monthly production API costs in USD (US Dollar).

Save 21% with Qwen: Qwen3.5-27B

Monthly Volume50M tokens/mo

Quick Presets:

Token Ratio (In vs Out)70% In / 30% Out

RAG / Search (10% out)Coding / Chat (50% out)

Qwen2.5 Coder 32B Instruct$38.10 / mo

In: $23.1Out: $15

Qwen: Qwen3.5-27B$30.23 / mo

In: $6.83Out: $23.4

Estimated Cost Delta

Qwen: Qwen3.5-27B is estimated to save $7.88/month ($95/year).

Based on 50M tokens (30% gen)

Key Selection Recommendations

1
Qwen2.5 Coder 32B Instruct:Pick Qwen2.5 Coder 32B Instruct when you are optimizing output cost.

Recommended Workload Routing

Repo / coding agentsinsufficient data

Need SWE-bench on both sides.

High-volume chatQwen2.5 Coder 32B Instruct

Lower output list price ($1/1M).

Voice / low-latency UIinsufficient data

Need TTFT on both sides.

Long-document RAGQwen: Qwen3.5-27B

Larger window (262k).

Screenshots / visioninsufficient data

Neither side is marked as vision-capable.

Target Workload	Recommended Pick	Evaluation Rationale
Repo / coding agents	insufficient data	Need SWE-bench on both sides.
High-volume chat	Qwen2.5 Coder 32B Instruct	Lower output list price ($1/1M).
Voice / low-latency UI	insufficient data	Need TTFT on both sides.
Long-document RAG	Qwen: Qwen3.5-27B	Larger window (262k).
Screenshots / vision	insufficient data	Neither side is marked as vision-capable.

Cast Your Matchup Vote

0 total votes

Community Discussions (0)

No comments posted on this matchup yet. Be the first to share an evaluation note!

Social Share Card Preview

Dynamic OpenGraph banner generated at /compare/qwen-2-5-coder-32b-instruct-vs-qwen3-5-27b/opengraph-image

Save PNG

CompareLLM AI Matrix

Verified Head-to-Head Telemetry

LIVE BENCHMARK

Alibaba

Qwen2.5 Coder 32B Instruct

Elo 1,425

SWE-bench65.2%

Speed (tok/s)33k

Input / 1M Tokens$0.66/1M

Output / 1M Tokens$1/1M

Alibaba

Qwen: Qwen3.5-27B

Elo n/a

SWE-bench—

Speed (tok/s)262k

Input / 1M Tokens$0.195/1M

Output / 1M Tokens$1.56/1M

Frequently Asked Questions

Which is better overall, Qwen2.5 Coder 32B Instruct or Qwen: Qwen3.5-27B?: We do not have preference Elo for both models, so we do not declare an overall winner. Compare the metrics that exist.
Which is better at coding, Qwen2.5 Coder 32B Instruct or Qwen: Qwen3.5-27B?: SWE-bench is missing for at least one model, so we do not rank coding from a single number.
Which is cheaper to run in production?: Qwen2.5 Coder 32B Instruct output tokens are $1/1M versus $1.56/1M. Input prices and retry rates still move the real bill.
Which is faster, Qwen2.5 Coder 32B Instruct or Qwen: Qwen3.5-27B?: We do not have TTFT for both models.
Which has the larger context window?: Qwen: Qwen3.5-27B accepts 262k tokens versus 33k.
Can I self-host either model?: Neither model is marked open-weights here. You are comparing hosted APIs.
Are these scores from CompareLLM’s own evals?: No. V1 aggregates public leaderboards (and optional first-party latency pings). Each cell names its source and date. Read /methodology.
What does preference Elo mean on this page?: Preference Elo is a crowd vote from LMArena / Arena. People see two hidden answers and pick the one they like more. The model that wins more often gets a higher Elo. That means people preferred it — not that it passed a school test. It is not SWE-bench, not accuracy, and not a number we invent. Qwen2.5 Coder 32B Instruct and Qwen: Qwen3.5-27B Elo cells are dated snapshots from the named source — we do not compute Arena ourselves.

Similar Head-to-Head Comparisons

All Qwen2.5 Coder 32B Instruct matchups →|All Qwen: Qwen3.5-27B matchups →

Qwen2.5 Coder 32B Instruct vs Qwen 2.5 Plus (next qwen)Qwen 3 Max vs Qwen2.5 Coder 32B Instruct Qwen QwQ 32B vs Qwen2.5 Coder 32B Instruct Qwen 2.5 Coder 32B vs Qwen2.5 Coder 32B Instruct Qwen3 235B vs Qwen2.5 Coder 32B Instruct Qwen3.8 27B vs Qwen2.5 Coder 32B Instruct

Capability Percentiles

Relative percentile scores computed across all active models in the benchmark catalog.

Multi-Dimensional Capability Radar

Leaders Matchup Capability Radar

Comparing top standard models across 6 key capability dimensions. Tap any spoke or dot to inspect.

Tap any node to inspect

0–100 %ile

Side-by-Side Benchmark Matrix

Qwen2.5 Coder 32B InstructQwen: Qwen3.5-27B

Preference Elo

Qwen2.5 Coder 32B Instruct1,425

Qwen: Qwen3.5-27B—

Coding Elo

Qwen2.5 Coder 32B Instruct1,495

Qwen: Qwen3.5-27B—

LiveBench

Qwen2.5 Coder 32B Instruct64.8%

Qwen: Qwen3.5-27B—

SWE-bench

Qwen2.5 Coder 32B Instruct65.2%

Qwen: Qwen3.5-27B—

Input priceQwen: Qwen3.5-27B +$0.465/1M

Qwen2.5 Coder 32B Instruct$0.66/1M

Qwen: Qwen3.5-27B$0.195/1M

Output priceQwen2.5 Coder 32B Instruct +$0.56/1M

Qwen2.5 Coder 32B Instruct$1/1M

Qwen: Qwen3.5-27B$1.56/1M

Context windowQwen: Qwen3.5-27B +229k

Qwen2.5 Coder 32B Instruct33k

Qwen: Qwen3.5-27B262k

Benchmark

Qwen2.5 Coder 32B Instruct

Qwen: Qwen3.5-27B

Advantage Delta

Preference Elo

1,425

lmarena · Aug 17, 2026

—

Coding Elo

1,495

lmarena · Aug 17, 2026

—

LiveBench

64.8%

livebench · Aug 17, 2026

—

SWE-bench

65.2%

swebench · Aug 17, 2026

—

Input price

$0.66/1M

openrouter · Aug 17, 2026

$0.195/1M

openrouter · Aug 17, 2026

+$0.465/1M

Output price

$1/1M

openrouter · Aug 17, 2026

$1.56/1M

openrouter · Aug 17, 2026

+$0.56/1M

Context window

33k

openrouter · Aug 17, 2026

262k

openrouter · Aug 17, 2026

+229k

Recommended Workload Routing

Repo / coding agentsinsufficient data

Need SWE-bench on both sides.

High-volume chatQwen2.5 Coder 32B Instruct

Lower output list price ($1/1M).

Voice / low-latency UIinsufficient data

Need TTFT on both sides.

Long-document RAGQwen: Qwen3.5-27B

Larger window (262k).

Screenshots / visioninsufficient data

Neither side is marked as vision-capable.

Target Workload	Recommended Pick	Evaluation Rationale
Repo / coding agents	insufficient data	Need SWE-bench on both sides.
High-volume chat	Qwen2.5 Coder 32B Instruct	Lower output list price ($1/1M).
Voice / low-latency UI	insufficient data	Need TTFT on both sides.
Long-document RAG	Qwen: Qwen3.5-27B	Larger window (262k).
Screenshots / vision	insufficient data	Neither side is marked as vision-capable.

Frequently Asked Questions

Which is better overall, Qwen2.5 Coder 32B Instruct or Qwen: Qwen3.5-27B?

We do not have preference Elo for both models, so we do not declare an overall winner. Compare the metrics that exist.

Which is better at coding, Qwen2.5 Coder 32B Instruct or Qwen: Qwen3.5-27B?

SWE-bench is missing for at least one model, so we do not rank coding from a single number.

Which is cheaper to run in production?

Qwen2.5 Coder 32B Instruct output tokens are $1/1M versus $1.56/1M. Input prices and retry rates still move the real bill.

Which is faster, Qwen2.5 Coder 32B Instruct or Qwen: Qwen3.5-27B?

We do not have TTFT for both models.

Which has the larger context window?

Qwen: Qwen3.5-27B accepts 262k tokens versus 33k.

Can I self-host either model?

Neither model is marked open-weights here. You are comparing hosted APIs.

Are these scores from CompareLLM’s own evals?

No. V1 aggregates public leaderboards (and optional first-party latency pings). Each cell names its source and date. Read /methodology.

What does preference Elo mean on this page?

Preference Elo is a crowd vote from LMArena / Arena. People see two hidden answers and pick the one they like more. The model that wins more often gets a higher Elo. That means people preferred it — not that it passed a school test. It is not SWE-bench, not accuracy, and not a number we invent. Qwen2.5 Coder 32B Instruct and Qwen: Qwen3.5-27B Elo cells are dated snapshots from the named source — we do not compute Arena ourselves.