Compare · ModelsLive · 2 picked · head to head
Baichuan 2-7B vs Qwen2.5 Coder 1.5B Instruct
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Baichuan 2-7B wins on 2/4 benchmarks
Baichuan 2-7B wins 2 of 4 shared benchmarks. Leads in knowledge.
Category leads
knowledge·Baichuan 2-7Bmath·Qwen2.5 Coder 1.5B Instruct
Hype vs Reality
Attention vs performance
Baichuan 2-7B
#186 by perf·no signal
Qwen2.5 Coder 1.5B Instruct
#198 by perf·#2 by attention
Best value
Pricing unknown
Baichuan 2-7B
n/a
no price
Qwen2.5 Coder 1.5B Instruct
n/a
no price
Vendor risk
Who is behind the model
U
Unknown
private · undisclosed
Alibaba (Qwen)
$293.0B·Tier 1
Head to head
4 benchmarks · 2 models
Baichuan 2-7BQwen2.5 Coder 1.5B Instruct
ARC AI2
Qwen2.5 Coder 1.5B Instruct leads by +16.9
AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval.
Baichuan 2-7B
10.0
Qwen2.5 Coder 1.5B Instruct
26.9
GSM8K
Qwen2.5 Coder 1.5B Instruct leads by +41.2
Grade School Math 8K · 8,500 linguistically diverse grade-school math word problems that require multi-step reasoning to solve.
Baichuan 2-7B
24.6
Qwen2.5 Coder 1.5B Instruct
65.8
HellaSwag
Baichuan 2-7B leads by +8.3
HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios.
Baichuan 2-7B
57.3
Qwen2.5 Coder 1.5B Instruct
49.1
MMLU
Baichuan 2-7B leads by +0.8
Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge.
Baichuan 2-7B
38.9
Qwen2.5 Coder 1.5B Instruct
38.1
Full benchmark table
| Benchmark | Baichuan 2-7B | Qwen2.5 Coder 1.5B Instruct |
|---|---|---|
ARC AI2 AI2 Reasoning Challenge · tests grade-school level science knowledge with multiple-choice questions requiring reasoning beyond simple retrieval. | 10.0 | 26.9 |
GSM8K Grade School Math 8K · 8,500 linguistically diverse grade-school math word problems that require multi-step reasoning to solve. | 24.6 | 65.8 |
HellaSwag HellaSwag · tests commonsense reasoning by asking models to predict the most plausible continuation of everyday scenarios. | 57.3 | 49.1 |
MMLU Massive Multitask Language Understanding · 57 subjects spanning STEM, humanities, social sciences, and more. The standard benchmark for broad knowledge. | 38.9 | 38.1 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
U Baichuan 2-7B | — | — | — | — |
| — | — | — | — |