Qwen2.5 Coder 32B vs StarCoder2 15B

What the listed data says

Qwen2.5 Coder 32B: 128K tokens; StarCoder2 15B: 16K tokens. Qwen2.5 Coder 32B has the larger listed context window. Qwen2.5 Coder 32B is listed for coding, agentic workflows, long-context tasks; StarCoder2 15B is listed for coding. Shared catalog benchmark entries: HumanEval: Qwen2.5 Coder 32B 92.7%; StarCoder2 15B 46.5%; MBPP: Qwen2.5 Coder 32B 90.2%; StarCoder2 15B 65.2%. These records are reference data, not a controlled head-to-head test; verify current pricing and specifications with each provider.

Catalog source links

Use these model and provider links to verify current specifications, pricing, and availability.

Computed differences from catalog data

The listed context windows are 128K tokens and 16K tokens; the larger is 8× the smaller.
On shared numeric benchmark entries, HumanEval: Qwen2.5 Coder 32B − StarCoder2 15B = +46.2 percentage points (92.7% vs 46.5%); MBPP: Qwen2.5 Coder 32B − StarCoder2 15B = +25 percentage points (90.2% vs 65.2%).
The first listed parameter quantities are 32B and 15.5B; the larger quantity is 2.06× the smaller. Total and active MoE parameters are not directly equivalent.
The catalog release dates are November 2024 and February 2024, about 9 months apart at the precision supplied.

Cost example: 1M input and 1M output tokens

A cost example for 1M input plus 1M output tokens cannot be calculated reliably because separate numeric input and output rates are not listed for both models.

Choose Qwen2.5 Coder 32B if…

Choose Qwen2.5 Coder 32B when your task aligns with catalog-listed uses or strengths such as Full-stack development, Code completion and review, Debugging complex code, Software architecture.

Choose StarCoder2 15B if…

Choose StarCoder2 15B when your task aligns with catalog-listed uses or strengths such as IDE completion in rare or niche programming languages, Organizations prioritizing ethical AI data sourcing, Multi-language polyglot projects spanning 600+ languages, 600+ programming languages — broadest language coverage of any open code model.

Qwen2.5 Coder 32B

Browse Qwen models

State-of-the-art code model from Qwen2.5 — leads HumanEval and coding benchmarks for 32B class.

Parameters: 32B

Context: 128K tokens

Catalog pricing: Free via LITHOVEX

Qwen Coding · Agentic · Long Context

View Qwen2.5 Coder 32B model profile

Try Qwen2.5 Coder 32B Free

B
StarCoder2 15B

Browse bigcode models

BigCode's StarCoder2 15B — trained on the Stack v2 with 600+ programming languages. Great for code completion and repository-level tasks.

Parameters: 15.5B

Context: 16K tokens

Catalog pricing: Apache 2.0 / BigCode OpenRAIL — free

bigcode Coding Open Source

View StarCoder2 15B model profile

Try StarCoder2 15B Free

Feature Comparison — Qwen2.5 Coder 32B vs StarCoder2 15B

Catalog specifications for Qwen2.5 Coder 32B and StarCoder2 15B
Specification	Qwen2.5 Coder 32B	StarCoder2 15B
Provider	Qwen	bigcode
Parameters	32B	15.5B
Context window	128K tokens	16K tokens
Maximum output	Not listed	Not listed
Catalog pricing	Free via LITHOVEX	Apache 2.0 / BigCode OpenRAIL — free
Listed input/output cost per 1M tokens	input $0.3, output $0.7 per 1M tokens	Separate numeric input/output rates not listed
License	Apache 2.0	BigCode Open RAIL-M v1
Coding support	Listed	Listed
Reasoning support	Not listed	Not listed
Vision input	Not listed	Not listed
Multimodal	Not listed	Not listed
Open weights	Not listed	Yes
Benchmark: HumanEval	92.7%	46.5%
Benchmark: MBPP	90.2%	65.2%

Figures are transcribed from LITHOVEX model records. Benchmark versions and evaluation setups may differ; pricing and specifications can change. Check each provider's current documentation before making a decision.

Frequently Asked Questions

What is different between Qwen2.5 Coder 32B and StarCoder2 15B?

Qwen2.5 Coder 32B: 128K tokens; StarCoder2 15B: 16K tokens. Qwen2.5 Coder 32B has the larger listed context window. Qwen2.5 Coder 32B is listed for coding, agentic workflows, long-context tasks; StarCoder2 15B is listed for coding. Shared catalog benchmark entries: HumanEval: Qwen2.5 Coder 32B 92.7%; StarCoder2 15B 46.5%; MBPP: Qwen2.5 Coder 32B 90.2%; StarCoder2 15B 65.2%. These records are reference data, not a controlled head-to-head test; verify current pricing and specifications with each provider.

What benchmark values are listed for Qwen2.5 Coder 32B and StarCoder2 15B?

HumanEval: Qwen2.5 Coder 32B 92.7%; StarCoder2 15B 46.5%; MBPP: Qwen2.5 Coder 32B 90.2%; StarCoder2 15B 65.2%. The catalog values may use different versions and evaluation setups, so they are reference points rather than a controlled head-to-head result.

Which model has the larger listed context window?

Qwen2.5 Coder 32B: 128K tokens; StarCoder2 15B: 16K tokens. Qwen2.5 Coder 32B has the larger listed context window.

What are the listed costs per million tokens for Qwen2.5 Coder 32B and StarCoder2 15B?

Qwen2.5 Coder 32B: input $0.3, output $0.7 per 1M tokens (Free via LITHOVEX). StarCoder2 15B: Separate numeric input/output rates not listed (Apache 2.0 / BigCode OpenRAIL — free). Where separate numeric input and output rates are available, the worked example uses 1M input plus 1M output tokens. Confirm current terms with each provider.

How far apart are the listed release dates for Qwen2.5 Coder 32B and StarCoder2 15B?

The catalog lists Qwen2.5 Coder 32B as November 2024 and StarCoder2 15B as February 2024, about 9 months apart at the precision supplied.

When was the comparison data last updated?

The model catalog records 2026-10-16 as its data update date. Provider specifications and prices may have changed since then.

How should I choose between Qwen2.5 Coder 32B and StarCoder2 15B?

Compare the listed specifications and capabilities, review the benchmark notes, then test both with the same representative prompt for your task. This page does not claim a universal winner.

Explore these models and related comparisons

Read the Qwen2.5 Coder 32B model profile and StarCoder2 15B model profile, or browse the model directory.

Return to the comparison directory