# Inkling vs Qwen3-Coder 480B-A35B Instruct

> Inkling is the stronger model overall, scoring 44.1 to 37.7 on the Noometry Index.

- Canonical page: https://noometry.com/compare/inkling-vs-qwen3-coder-480b-a35b-instruct
- Last updated: 2026-10-10
- Shared benchmarks: 20

## Summary

- They share 20 benchmarks with published results for both. Inkling scores higher in 7 categories and Qwen3-Coder 480B-A35B Instruct in 2 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Inkling leads 55.2 to 36.4.
- The biggest single-benchmark swing is WeirdML: 32.3% for Inkling and 41.2% for Qwen3-Coder 480B-A35B Instruct.
- Inkling is cheaper at $1.87 / $4.68 per million input/output tokens, against $1.50 / $7.50 for Qwen3-Coder 480B-A35B Instruct.
- Qwen3-Coder 480B-A35B Instruct accepts more context: 262K tokens versus 66K.

## Snapshot

| | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| Provider | Thinking Machines Lab | Alibaba (Qwen) |
| Noometry Index | 44.1 | 37.7 |
| Rank | 75 | 186 |
| Context | 66K | 262K |
| Input $/M | $1.87 | $1.50 |
| Output $/M | $4.68 | $7.50 |
| Weights | Open | Open |

## Coding

- Inkling: 34.1 (#229)
- Qwen3-Coder 480B-A35B Instruct: 36.2 (#199)

| Benchmark | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena WebDev | 1412 | 1275 |
| WeirdML | 32.3% | 41.2% |
| LMArena Coding | 1463 | 1413 |
| ALE-Bench | 946 | 461.45 |
| FrontierCode | 14% | — |
| SWE-bench Verified (bash only) | — | 55.4% |
| FrontierSWE | 4.1% | — |
| SciCode | 47% | — |
| GSO | — | 4.9% |
| AlgoTune | — | 1.44 |

## Agentic & Tool Use

- Inkling: 29.5 (#76)
- Qwen3-Coder 480B-A35B Instruct: 23.7 (#116)

| Benchmark | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| Terminal-Bench | — | 27.2% |
| APEX-Agents | 33.8% | — |
| τ²-bench Banking | 25% | — |

## Reasoning

- Inkling: 40.6 (#48)
- Qwen3-Coder 480B-A35B Instruct: 24.7 (#141)

| Benchmark | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1452 | 1372 |
| ARC-AGI-2 | 36.5% | — |
| SimpleBench | 50% | — |
| Kagi LLM Benchmark | — | 49.5% |
| ARC-AGI-1 | 79.5% | — |
| CritPt | 5.4% | — |
| Chess Puzzles | 21% | — |
| DTBench | 87.5% | — |
| LMCA | 37.6% | — |
| Epoch Capabilities Index | 148.6 | — |

## Math

- Inkling: 31.2 (#198)
- Qwen3-Coder 480B-A35B Instruct: 34.9 (#151)

| Benchmark | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Math | 1482 | 1364 |
| FrontierMath (Tiers 1-3) | 33.3% | — |
| FrontierMath Tier 4 | 4.9% | — |
| OTIS Mock AIME 2024-2025 | 88.9% | — |
| ProofBench | 0% | — |

## Knowledge

- Inkling: 55.2 (#44)
- Qwen3-Coder 480B-A35B Instruct: 36.4 (#159)

| Benchmark | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Expert | 1468 | 1337 |
| GPQA Diamond | 88.3% | — |
| SimpleQA Verified | 40.3% | — |

## Multilingual

- Inkling: 54.2 (#49)
- Qwen3-Coder 480B-A35B Instruct: 47.8 (#143)

| Benchmark | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Non-English | 1434 | 1345 |
| LMArena Chinese | 1495 | 1356 |
| LMArena French | 1453 | 1394 |
| LMArena German | 1456 | 1324 |
| LMArena Japanese | 1428 | 1309 |
| LMArena Korean | 1401 | 1307 |
| LMArena Russian | 1430 | 1367 |
| LMArena Spanish | 1442 | 1360 |

## Instruction Following

- Inkling: 75.3 (#65)
- Qwen3-Coder 480B-A35B Instruct: 71.8 (#142)

| Benchmark | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Instruction Following | 1427 | 1356 |

## Long Context

- Inkling: 43.9 (#79)
- Qwen3-Coder 480B-A35B Instruct: 42.0 (#126)

| Benchmark | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Longer Query | 1436 | 1379 |

## Writing & Preference

- Inkling: 65.5 (#48)
- Qwen3-Coder 480B-A35B Instruct: 55.5 (#142)

| Benchmark | Inkling | Qwen3-Coder 480B-A35B Instruct |
|---|---|---|
| LMArena Text | 1443 | 1356 |
| LMArena Creative Writing | 1390 | 1333 |
| LMArena Multi-Turn | 1436 | 1365 |
| EQ-Bench Creative Writing | 1611 | — |
| EQ-Bench 4 | 1226 | — |

## FAQ

### Is Inkling better than Qwen3-Coder 480B-A35B Instruct?

Inkling is the stronger model overall, scoring 44.1 to 37.7 on the Noometry Index.

### Which is cheaper, Inkling or Qwen3-Coder 480B-A35B Instruct?

Inkling is cheaper. It lists at $1.87 per million input tokens and $4.68 per million output tokens; Qwen3-Coder 480B-A35B Instruct lists at $1.50 and $7.50.

### Is Inkling or Qwen3-Coder 480B-A35B Instruct better for coding?

Qwen3-Coder 480B-A35B Instruct scores higher on coding benchmarks: 36.2 versus 34.1 in the Noometry coding category.

### Which has the bigger context window?

Qwen3-Coder 480B-A35B Instruct does, with 262K tokens against 66K.

### How many benchmarks do Inkling and Qwen3-Coder 480B-A35B Instruct share?

20 benchmarks have published results for both models. Inkling has 41 scored results on Noometry and Qwen3-Coder 480B-A35B Instruct has 25.