Model comparison
Inkling vs Qwen2.5 32B Instruct
Inkling is the stronger model overall, scoring 44.1 to 29.7 on the Noometry Index. Qwen2.5 32B Instruct costs 2.1× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Last verified . 4 shared benchmarks.
Alibaba (Qwen)Summary
- They share 4 benchmarks with published results for both. Inkling scores higher in 3 categories and Qwen2.5 32B Instruct in 1 category; 4 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Inkling leads 55.2 to 24.9.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 88.9% for Inkling and 7.4% for Qwen2.5 32B Instruct.
- Qwen2.5 32B Instruct is cheaper at $0.70 / $2.80 per million input/output tokens, against $1.87 / $4.68 for Inkling.
- Qwen2.5 32B Instruct accepts more context: 131K tokens versus 66K.
Side by side
| Inkling | Qwen2.5 32B Instruct | |
|---|---|---|
| Provider | Thinking Machines Lab | Alibaba (Qwen) |
| Noometry Index | 44.1 | 29.7 |
| Released | 2026-07-15 | 2024-09 |
| Weights | Open | Open |
| Context window | 66K | 131K |
| Max output | 66K | 8K |
| Input $ / M tokens | $1.87 | $0.70 |
| Output $ / M tokens | $4.68 | $2.80 |
| Results tracked | 41 | 7 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen2.5 32B Instruct leads
Inkling: 34.1 (#229), Qwen2.5 32B Instruct: 38.1 (#165)
| Benchmark | Inkling | Qwen2.5 32B Instruct |
|---|---|---|
| FrontierCode | 14% | — |
| LMArena WebDev | 1412 | — |
| FrontierSWE | 4.1% | — |
| SciCode | 47% | — |
| WeirdML | 32.3% | — |
| BigCodeBench Instruct | — | 45% |
| LMArena Coding | 1463 | — |
| BigCodeBench Complete | — | 52.3% |
| ALE-Bench | 946 | — |
Agentic & Tool Use Not comparable
Inkling: 29.5 (#76), Qwen2.5 32B Instruct: —
| Benchmark | Inkling | Qwen2.5 32B Instruct |
|---|---|---|
| APEX-Agents | 33.8% | — |
| τ²-bench Banking | 25% | — |
Reasoning Inkling leads
Inkling: 40.6 (#48), Qwen2.5 32B Instruct: 18.5 (#262)
| Benchmark | Inkling | Qwen2.5 32B Instruct |
|---|---|---|
| Chess Puzzles | 21% | 0% |
| Epoch Capabilities Index | 148.6 | 128.52 |
| ARC-AGI-2 | 36.5% | — |
| SimpleBench | 50% | — |
| ARC-AGI-1 | 79.5% | — |
| CritPt | 5.4% | — |
| LMArena Hard Prompts | 1452 | — |
| DTBench | 87.5% | — |
| LMCA | 37.6% | — |
Math Inkling leads
Inkling: 31.2 (#198), Qwen2.5 32B Instruct: 15.3 (#289)
| Benchmark | Inkling | Qwen2.5 32B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 7.4% |
| FrontierMath (Tiers 1-3) | 33.3% | — |
| FrontierMath Tier 4 | 4.9% | — |
| ProofBench | 0% | — |
| LMArena Math | 1482 | — |
| MATH Level 5 | — | 56.1% |
Knowledge Inkling leads
Inkling: 55.2 (#44), Qwen2.5 32B Instruct: 24.9 (#259)
| Benchmark | Inkling | Qwen2.5 32B Instruct |
|---|---|---|
| GPQA Diamond | 88.3% | 46.1% |
| SimpleQA Verified | 40.3% | — |
| LMArena Expert | 1468 | — |
Multilingual Not comparable
Inkling: 54.2 (#49), Qwen2.5 32B Instruct: —
| Benchmark | Inkling | Qwen2.5 32B Instruct |
|---|---|---|
| LMArena Non-English | 1434 | — |
| LMArena Chinese | 1495 | — |
| LMArena French | 1453 | — |
| LMArena German | 1456 | — |
| LMArena Japanese | 1428 | — |
| LMArena Korean | 1401 | — |
| LMArena Russian | 1430 | — |
| LMArena Spanish | 1442 | — |
Instruction Following Not comparable
Inkling: 75.3 (#65), Qwen2.5 32B Instruct: —
| Benchmark | Inkling | Qwen2.5 32B Instruct |
|---|---|---|
| LMArena Instruction Following | 1427 | — |
Long Context Not comparable
Inkling: 43.9 (#79), Qwen2.5 32B Instruct: —
| Benchmark | Inkling | Qwen2.5 32B Instruct |
|---|---|---|
| LMArena Longer Query | 1436 | — |
Writing & Preference Not comparable
Inkling: 65.5 (#48), Qwen2.5 32B Instruct: —
| Benchmark | Inkling | Qwen2.5 32B Instruct |
|---|---|---|
| LMArena Text | 1443 | — |
| LMArena Creative Writing | 1390 | — |
| EQ-Bench Creative Writing | 1611 | — |
| EQ-Bench 4 | 1226 | — |
| LMArena Multi-Turn | 1436 | — |
Frequently asked questions
Is Inkling better than Qwen2.5 32B Instruct?
Inkling is the stronger model overall, scoring 44.1 to 29.7 on the Noometry Index. Qwen2.5 32B Instruct costs 2.1× less per token, which makes it the better buy when Inkling's lead doesn't matter for your workload.
Which is cheaper, Inkling or Qwen2.5 32B Instruct?
Qwen2.5 32B Instruct is cheaper. It lists at $0.70 per million input tokens and $2.80 per million output tokens; Inkling lists at $1.87 and $4.68.
Is Inkling or Qwen2.5 32B Instruct better for coding?
Qwen2.5 32B Instruct scores higher on coding benchmarks: 38.1 versus 34.1 in the Noometry coding category.
Which has the bigger context window?
Qwen2.5 32B Instruct does, with 131K tokens against 66K.
How many benchmarks do Inkling and Qwen2.5 32B Instruct share?
4 benchmarks have published results for both models. Inkling has 41 scored results on Noometry and Qwen2.5 32B Instruct has 7.