Compare · ModelsLive · 2 picked · head to head
StarCoder 2 15B vs Gemma 2B
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Gemma 2B wins on 6/10 benchmarks
Gemma 2B wins 6 of 10 shared benchmarks. Leads in knowledge · general · reasoning.
Category leads
knowledge·Gemma 2Bmath·StarCoder 2 15Bgeneral·Gemma 2Blanguage·StarCoder 2 15Breasoning·Gemma 2B
Hype vs Reality
Attention vs performance
StarCoder 2 15B
#204 by perf·no signal
Gemma 2B
#189 by perf·no signal
Vendor risk
Who is behind the model
U
Unknown
private · undisclosed
Google DeepMind
$4.00T·Tier 1
Head to head
10 benchmarks · 2 models
StarCoder 2 15BGemma 2B
ARC AI2
StarCoder 2 15B leads by +6.8
AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval.
StarCoder 2 15B
29.6
Gemma 2B
22.8
GSM8K
StarCoder 2 15B leads by +40.0
Grade School Math 8K · 8,500 linguistically diverse grade-school math word problems that require multi-step reasoning to solve.
StarCoder 2 15B
57.7
Gemma 2B
17.7
BBH (HuggingFace)
Gemma 2B leads by +0.8
StarCoder 2 15B
20.4
Gemma 2B
21.1
GPQA
Gemma 2B leads by +1.8
StarCoder 2 15B
3.1
Gemma 2B
4.9
IFEval
StarCoder 2 15B leads by +1.2
StarCoder 2 15B
27.8
Gemma 2B
26.6
MATH Level 5
Gemma 2B leads by +1.4
StarCoder 2 15B
6.0
Gemma 2B
7.4
MMLU-PRO
Gemma 2B leads by +6.6
StarCoder 2 15B
15.0
Gemma 2B
21.6
MUSR
Gemma 2B leads by +8.1
StarCoder 2 15B
2.9
Gemma 2B
11.0
MMLU
StarCoder 2 15B leads by +29.1
Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge.
StarCoder 2 15B
52.1
Gemma 2B
23.1
Winogrande
Gemma 2B leads by +2.2
WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs.
StarCoder 2 15B
28.6
Gemma 2B
30.8
Full benchmark table
| Benchmark | StarCoder 2 15B | Gemma 2B |
|---|---|---|
ARC AI2 AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval. | 29.6 | 22.8 |
GSM8K Grade School Math 8K · 8,500 linguistically diverse grade-school math word problems that require multi-step reasoning to solve. | 57.7 | 17.7 |
BBH (HuggingFace) | 20.4 | 21.1 |
GPQA | 3.1 | 4.9 |
IFEval | 27.8 | 26.6 |
MATH Level 5 | 6.0 | 7.4 |
MMLU-PRO | 15.0 | 21.6 |
MUSR | 2.9 | 11.0 |
MMLU Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge. | 52.1 | 23.1 |
Winogrande WinoGrande · large-scale commonsense reasoning benchmark where models must resolve ambiguous pronouns in carefully constructed sentence pairs. | 28.6 | 30.8 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
U StarCoder 2 15B | — | — | — | — |
| — | — | — | — |