Model comparison
Qwen2.5-Coder-32B vs Sonar
Sonar is the stronger model overall, scoring 38.5 to 33.4 on the Noometry Index.
Last verified . 7 shared benchmarks.
Alibaba (Qwen)
PerplexitySummary
- They share 7 benchmarks with published results for both. Qwen2.5-Coder-32B scores higher in 1 category and Sonar in 4 categories; 3 gaps are clear of the uncertainty.
- The widest gap is in coding, where Sonar leads 35.7 to 22.6.
- The biggest single-benchmark swing is LiveBench Coding: 56.9% for Qwen2.5-Coder-32B and 35.1% for Sonar.
- Qwen2.5-Coder-32B is cheaper at $0.66 / $1 per million input/output tokens, against $1 / $1 for Sonar.
- Sonar accepts more context: 128K tokens versus 33K.
- Qwen2.5-Coder-32B has downloadable open weights; the other is API-only.
Side by side
| Qwen2.5-Coder-32B | Sonar | |
|---|---|---|
| Provider | Alibaba (Qwen) | Perplexity |
| Noometry Index | 33.4 | 38.5 |
| Released | 2024-09-18 | 2024-01-01 |
| Weights | Open | Proprietary |
| Context window | 33K | 128K |
| Max output | 29K | 4K |
| Input $ / M tokens | $0.66 | $1 |
| Output $ / M tokens | $1 | $1 |
| Results tracked | 31 | 7 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Sonar leads
Qwen2.5-Coder-32B: 22.6 (#333), Sonar: 35.7 (#221)
| Benchmark | Qwen2.5-Coder-32B | Sonar |
|---|---|---|
| LiveBench Coding | 56.9% | 35.1% |
| SWE-bench Verified (bash only) | 9% | — |
| Aider Polyglot | 16.4% | — |
| BigCodeBench Instruct | 49% | — |
| LMArena Coding | 1276 | — |
| BigCodeBench Complete | 58% | — |
| HumanEval+ | 87.2% | — |
| MBPP+ | 77% | — |
Reasoning Too close to call
Qwen2.5-Coder-32B: 21.2 (#225), Sonar: 21.1 (#227)
| Benchmark | Qwen2.5-Coder-32B | Sonar |
|---|---|---|
| LiveBench Reasoning | 42.1% | 46.3% |
| LiveBench Data Analysis | 49.9% | 37.9% |
| LiveBench | 46.2% | 46.9% |
| LMArena Hard Prompts | 1251 | — |
| Epoch Capabilities Index | 119.49 | — |
| HellaSwag | 83% | — |
| WinoGrande | 80.8% | — |
Math Too close to call
Qwen2.5-Coder-32B: 33.3 (#204), Sonar: 33.7 (#200)
| Benchmark | Qwen2.5-Coder-32B | Sonar |
|---|---|---|
| LiveBench Math | 46.6% | 41.6% |
| LMArena Math | 1251 | — |
| GSM8K | 93% | — |
Knowledge Not comparable
Qwen2.5-Coder-32B: 33.4 (#203), Sonar: —
| Benchmark | Qwen2.5-Coder-32B | Sonar |
|---|---|---|
| LMArena Expert | 1221 | — |
| ARC (AI2) Challenge | 70.5% | — |
| MMLU | 79.1% | — |
Multilingual Not comparable
Qwen2.5-Coder-32B: 37.8 (#235), Sonar: —
| Benchmark | Qwen2.5-Coder-32B | Sonar |
|---|---|---|
| LMArena Non-English | 1205 | — |
| LMArena Chinese | 1222 | — |
| LMArena Russian | 1228 | — |
Instruction Following Sonar leads
Qwen2.5-Coder-32B: 61.4 (#245), Sonar: 71.4 (#150)
| Benchmark | Qwen2.5-Coder-32B | Sonar |
|---|---|---|
| LiveBench Instruction Following | 58.7% | 76.2% |
| LMArena Instruction Following | 1223 | — |
Long Context Not comparable
Qwen2.5-Coder-32B: 38.0 (#208), Sonar: —
| Benchmark | Qwen2.5-Coder-32B | Sonar |
|---|---|---|
| LMArena Longer Query | 1251 | — |
Writing & Preference Sonar leads
Qwen2.5-Coder-32B: 41.6 (#240), Sonar: 52.6 (#167)
| Benchmark | Qwen2.5-Coder-32B | Sonar |
|---|---|---|
| LiveBench Language | 23.3% | 44.1% |
| LMArena Text | 1230 | — |
| LMArena Creative Writing | 1174 | — |
| LMArena Multi-Turn | 1222 | — |
Frequently asked questions
Is Qwen2.5-Coder-32B better than Sonar?
Sonar is the stronger model overall, scoring 38.5 to 33.4 on the Noometry Index.
Which is cheaper, Qwen2.5-Coder-32B or Sonar?
Qwen2.5-Coder-32B is cheaper. It lists at $0.66 per million input tokens and $1 per million output tokens; Sonar lists at $1 and $1.
Is Qwen2.5-Coder-32B or Sonar better for coding?
Sonar scores higher on coding benchmarks: 35.7 versus 22.6 in the Noometry coding category.
Which has the bigger context window?
Sonar does, with 128K tokens against 33K.
How many benchmarks do Qwen2.5-Coder-32B and Sonar share?
7 benchmarks have published results for both models. Qwen2.5-Coder-32B has 31 scored results on Noometry and Sonar has 7.