Noometry

Model comparison

Claude Haiku 4.5 vs Qwen2.5-Coder-32B

Claude Haiku 4.5 is the stronger model overall, scoring 39.3 to 32.9 on the Noometry Index. Qwen2.5-Coder-32B costs 2.7× less per token, which makes it the better buy when Claude Haiku 4.5's lead doesn't matter for your workload.

Last verified . 14 shared benchmarks.

Claude Haiku 4.5 Anthropic

39.3

Rank #155 Confirmed

Qwen2.5-Coder-32B Alibaba (Qwen)

32.9

Rank #239 Confirmed

Summary

  • They share 14 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 7 categories and Qwen2.5-Coder-32B in 1 category; 8 gaps are clear of the uncertainty.
  • The widest gap is in coding, where Claude Haiku 4.5 leads 43.9 to 22.4.
  • The biggest single-benchmark swing is SWE-bench Verified (bash only): 66.6% for Claude Haiku 4.5 and 9% for Qwen2.5-Coder-32B.
  • Qwen2.5-Coder-32B is cheaper at $0.66 / $1 per million input/output tokens, against $1 / $5 for Claude Haiku 4.5.
  • Claude Haiku 4.5 accepts more context: 200K tokens versus 33K.
  • Qwen2.5-Coder-32B has downloadable open weights; the other is API-only.

Side by side

Claude Haiku 4.5Qwen2.5-Coder-32B
ProviderAnthropicAlibaba (Qwen)
Noometry Index39.332.9
Released2025-10-152024-09-18
WeightsProprietaryOpen
Context window200K33K
Max output64K29K
Input $ / M tokens$1$0.66
Output $ / M tokens$5$1
Results tracked5331

Sponsored placements are available on pages like this one. Advertise on Noometry

Category by category

Coding Claude Haiku 4.5 leads

Claude Haiku 4.5: 43.9 (#71), Qwen2.5-Coder-32B: 22.4 (#328)

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
SWE-bench Verified (bash only)66.6%9%
LMArena Coding14541276
Aider Polyglot—16.4%
LMArena WebDev1330—
SWE-bench Multilingual64.7%—
SciCode43.3%—
WeirdML45.4%—
BigCodeBench Instruct—49%
LiveBench Coding—56.9%
BigCodeBench Complete—58%
ALE-Bench653.48—
HumanEval+—87.2%
MBPP+—77%

Agentic & Tool Use Not comparable

Claude Haiku 4.5: 33.5 (#50), Qwen2.5-Coder-32B: —

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
Terminal-Bench35.5%—
Berkeley Function Calling Leaderboard68.7%—
DeepResearch Bench45.5%—
BALROG31.2%—
ExploitBench13.7%—
Vending-Bench 2458.89—

Reasoning Qwen2.5-Coder-32B leads

Claude Haiku 4.5: 15.1 (#310), Qwen2.5-Coder-32B: 20.4 (#220)

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
LMArena Hard Prompts14201251
Epoch Capabilities Index142.42119.41
ARC-AGI-24%—
NYT Connections (extended)14.3%—
ARC-AGI-147.7%—
CritPt0%—
Chess Puzzles8%—
LiveBench Reasoning—42.1%
DTBench73.6%—
LiveBench Data Analysis—49.9%
LMCA30.9%—
ForecastBench61.4—
HellaSwag—83%
LiveBench—46.2%
WinoGrande—80.8%

Math Claude Haiku 4.5 leads

Claude Haiku 4.5: 43.0 (#69), Qwen2.5-Coder-32B: 30.8 (#203)

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
LMArena Math13951251
OTIS Mock AIME 2024-202566.7%—
Omni-MATH56.1%—
LiveBench Math—46.6%
MATH Level 596.4%—
FrontierMath (Feb 2025 set)5.9%—
FrontierMath Tier 4 (v1)2.1%—
GSM8K—93%

Knowledge Claude Haiku 4.5 leads

Claude Haiku 4.5: 37.6 (#146), Qwen2.5-Coder-32B: 32.9 (#200)

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
LMArena Expert14431221
GPQA Diamond71.2%—
SimpleQA Verified13.2%—
MMLU-Pro77.7%—
Vectara Hallucination Rate9.8%—
GPQA (HELM)60.5%—
ARC (AI2) Challenge—70.5%
MMLU—79.1%

Multimodal Not comparable

Claude Haiku 4.5: 27.0 (#111), Qwen2.5-Coder-32B: —

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
Blueprint-Bench 20%—
LMArena Document1420—

Multilingual Claude Haiku 4.5 leads

Claude Haiku 4.5: 50.1 (#124), Qwen2.5-Coder-32B: 38.0 (#230)

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
LMArena Non-English13761205
LMArena Chinese14151222
LMArena Russian13801228
LMArena French1408—
LMArena German1377—
LMArena Japanese1339—
LMArena Korean1346—
LMArena Spanish1420—

Instruction Following Claude Haiku 4.5 leads

Claude Haiku 4.5: 71.6 (#144), Qwen2.5-Coder-32B: 61.6 (#240)

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
LMArena Instruction Following14141223
LiveBench Instruction Following—58.7%
IFEval80.1%—

Long Context Claude Haiku 4.5 leads

Claude Haiku 4.5: 43.6 (#86), Qwen2.5-Coder-32B: 37.9 (#204)

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
LMArena Longer Query14271251

Writing & Preference Claude Haiku 4.5 leads

Claude Haiku 4.5: 58.1 (#119), Qwen2.5-Coder-32B: 41.8 (#235)

BenchmarkClaude Haiku 4.5Qwen2.5-Coder-32B
LMArena Text13961230
LMArena Creative Writing13711174
LMArena Multi-Turn14091222
WildBench83.9%—
EQ-Bench 41064—
LiveBench Language—23.3%

Frequently asked questions

Is Claude Haiku 4.5 better than Qwen2.5-Coder-32B?

Claude Haiku 4.5 is the stronger model overall, scoring 39.3 to 32.9 on the Noometry Index. Qwen2.5-Coder-32B costs 2.7× less per token, which makes it the better buy when Claude Haiku 4.5's lead doesn't matter for your workload.

Which is cheaper, Claude Haiku 4.5 or Qwen2.5-Coder-32B?

Qwen2.5-Coder-32B is cheaper. It lists at $0.66 per million input tokens and $1 per million output tokens; Claude Haiku 4.5 lists at $1 and $5.

Is Claude Haiku 4.5 or Qwen2.5-Coder-32B better for coding?

Claude Haiku 4.5 scores higher on coding benchmarks: 43.9 versus 22.4 in the Noometry coding category.

Which has the bigger context window?

Claude Haiku 4.5 does, with 200K tokens against 33K.

How many benchmarks do Claude Haiku 4.5 and Qwen2.5-Coder-32B share?

14 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and Qwen2.5-Coder-32B has 31.

Related comparisons

Go deeper