Model comparison
Inkling vs Qwen2.5 72B Instruct
Inkling is the stronger model overall, scoring 44.1 to 31.9 on the Noometry Index.
Last verified . 23 shared benchmarks.
Alibaba (Qwen)Summary
- They share 23 benchmarks with published results for both. Inkling scores higher in 9 categories and Qwen2.5 72B Instruct in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Inkling leads 55.1 to 27.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 88.9% for Inkling and 8.1% for Qwen2.5 72B Instruct.
- Both cost about the same: $1.87 input and $4.68 output per million tokens.
- Qwen2.5 72B Instruct accepts more context: 131K tokens versus 66K.
Side by side
| Inkling | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | Thinking Machines Lab | Alibaba (Qwen) |
| Noometry Index | 44.1 | 31.9 |
| Released | 2026-07-15 | 2024-09 |
| Weights | Open | Open |
| Context window | 66K | 131K |
| Max output | 66K | 8K |
| Input $ / M tokens | $1.87 | $1.40 |
| Output $ / M tokens | $4.68 | $5.60 |
| Results tracked | 41 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Inkling leads
Inkling: 34.5 (#234), Qwen2.5 72B Instruct: 33.2 (#260)
| Benchmark | Inkling | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | 32.3% | 16% |
| LMArena Coding | 1464 | 1292 |
| FrontierCode | 14% | — |
| LMArena WebDev | 1413 | — |
| FrontierSWE | 4.1% | — |
| SciCode | 47% | — |
| BigCodeBench Instruct | — | 45.8% |
| BigCodeBench Complete | — | 55.9% |
| ALE-Bench | 946 | — |
Agentic & Tool Use Inkling leads
Inkling: 29.6 (#85), Qwen2.5 72B Instruct: 22.1 (#133)
| Benchmark | Inkling | Qwen2.5 72B Instruct |
|---|---|---|
| APEX-Agents | 33.8% | — |
| TheAgentCompany | — | 5.7% |
| τ²-bench Banking | 25% | — |
| BALROG | — | 16.2% |
| METR Time Horizons | — | 35.8% |
Reasoning Inkling leads
Inkling: 40.4 (#56), Qwen2.5 72B Instruct: 22.3 (#199)
| Benchmark | Inkling | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1451 | 1271 |
| DTBench | 87.5% | 62.9% |
| LMCA | 37.6% | 13.4% |
| Epoch Capabilities Index | 148.54 | 129 |
| ARC-AGI-2 | 36.5% | — |
| SimpleBench | 50% | — |
| ARC-AGI-1 | 79.5% | — |
| CritPt | 5.4% | — |
| Chess Puzzles | 21% | — |
| BIG-Bench Hard | — | 79.8% |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math Inkling leads
Inkling: 31.3 (#225), Qwen2.5 72B Instruct: 19.3 (#287)
| Benchmark | Inkling | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 8.1% |
| LMArena Math | 1479 | 1283 |
| FrontierMath (Tiers 1-3) | 33.3% | — |
| FrontierMath Tier 4 | 4.9% | — |
| ProofBench | 0% | — |
| Omni-MATH | — | 33% |
| MATH Level 5 | — | 63.2% |
Knowledge Inkling leads
Inkling: 55.1 (#49), Qwen2.5 72B Instruct: 27.0 (#253)
| Benchmark | Inkling | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 88.3% | 49.1% |
| LMArena Expert | 1465 | 1245 |
| SimpleQA Verified | 40.3% | — |
| MMLU-Pro | — | 63.1% |
| Confabulations | — | 19.1% |
| GPQA (HELM) | — | 42.6% |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multilingual Inkling leads
Inkling: 54.0 (#52), Qwen2.5 72B Instruct: 41.0 (#213)
| Benchmark | Inkling | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1434 | 1252 |
| LMArena Chinese | 1490 | 1272 |
| LMArena French | 1458 | 1280 |
| LMArena German | 1446 | 1234 |
| LMArena Japanese | 1429 | 1180 |
| LMArena Korean | 1404 | 1188 |
| LMArena Russian | 1429 | 1264 |
| LMArena Spanish | 1448 | 1256 |
Instruction Following Inkling leads
Inkling: 75.1 (#71), Qwen2.5 72B Instruct: 65.5 (#221)
| Benchmark | Inkling | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Instruction Following | 1426 | 1254 |
| IFEval | — | 80.6% |
Long Context Inkling leads
Inkling: 43.8 (#86), Qwen2.5 72B Instruct: 38.9 (#188)
| Benchmark | Inkling | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1434 | 1282 |
Writing & Preference Inkling leads
Inkling: 65.2 (#51), Qwen2.5 72B Instruct: 46.7 (#215)
| Benchmark | Inkling | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1441 | 1269 |
| LMArena Creative Writing | 1387 | 1221 |
| LMArena Multi-Turn | 1436 | 1272 |
| EQ-Bench Creative Writing | 1611 | — |
| WildBench | — | 80.2% |
| EQ-Bench 4 | 1226 | — |
Frequently asked questions
Is Inkling better than Qwen2.5 72B Instruct?
Inkling is the stronger model overall, scoring 44.1 to 31.9 on the Noometry Index.
Which is cheaper, Inkling or Qwen2.5 72B Instruct?
Qwen2.5 72B Instruct is cheaper. It lists at $1.40 per million input tokens and $5.60 per million output tokens; Inkling lists at $1.87 and $4.68.
Is Inkling or Qwen2.5 72B Instruct better for coding?
Inkling scores higher on coding benchmarks: 34.5 versus 33.2 in the Noometry coding category.
Which has the bigger context window?
Qwen2.5 72B Instruct does, with 131K tokens against 66K.
How many benchmarks do Inkling and Qwen2.5 72B Instruct share?
23 benchmarks have published results for both models. Inkling has 41 scored results on Noometry and Qwen2.5 72B Instruct has 43.