Inkling 2 Small vs Muse Spark Open
Both models appear in the same evaluation table on the official model card, which is the only place they are measured under identical conditions and reported by the same party.
Across 52 benchmarks where both models report a score, Inkling 2 Small is higher on 43 and Muse Spark Open is higher on 8 , with 1 exact ties. Inkling 2 Small has 594B total parameters with 17B active, and is priced at $1 per million output tokens on Tinker. Its largest margin is on AA-Briefcase v1.1. Its largest deficit is on AA-Omniscience v1.1.
Head to head
- Position
- Inkling 2 Small: the subject of this site · Muse Spark Open: open weights, larger class
- Bechmarks scored
- 52 shared rows, out of 71 total
- Record
- 43 wins · 8 losses · 1 ties
Largest differences
| Benchmark | Inkling 2 Small | Muse Spark Open | Gap |
|---|---|---|---|
| AA-Briefcase v1.1 · Knowledge Work | 1,527 | 1,279 | + 248 |
| GDPVal-AA v2.1 · Knowledge Work | 1,630 | 1,401 | + 229 |
| DesignArena Game Dev · Visual Creation | 1,428 | 1,270 | + 158 |
| DesignArena 3D Design · Visual Creation | 1,401 | 1,261 | + 140 |
| DesignArena Website · Visual Creation | 1,340 | 1,303 | + 37 |
| Apex 2025 · STEM Reasoning | 63.5% | 32.3% | + 31.2 |
| BabyVision · Visual Perception | 68.6% | 44.3% | + 24.3 |
| FrontierCode v2.0 · Coding | 25.4% | 7.6% | + 17.8 |
| Tax Agent Bench · Finance Work | 65.8% | 52.1% | + 13.7 |
| DeepSWE v1.1 · Coding | 72.1% | 60% | + 12.1 |
Every shared benchmark
| Benchmark | Inkling 2 Small | Muse Spark Open | Leader |
|---|---|---|---|
| Third Party Index | |||
| AA Index (v5.0) | 30.6 | 24.9 | Inkling 2 Small |
| Vals Index (v2.2) | 56.1% | 48.6% | Inkling 2 Small |
| Coding | |||
| FrontierCode v2.0 | 25.4% | 7.6% | Inkling 2 Small |
| DeepSWE v1.1 | 72.1% | 60% | Inkling 2 Small |
| Terminal Bench Science v0.1 | 5.7% | 1.9% | Inkling 2 Small |
| Terminal Bench v4.0 | 13.1% | 8.1% | Inkling 2 Small |
| Terminal Bench 2.1 · Inkling 2 Small: Pi harness | 85.1% | 76.4% | Inkling 2 Small |
| Terminal Bench v2.1 (Terminus) | 81.3% | 76.4% | Inkling 2 Small |
| Vibe Code Bench v1.1 | 81.6% | 76.2% | Inkling 2 Small |
| SWEBench Multilingual | 81.2% | 77.5% | Inkling 2 Small |
| SWEBench Pro v2 (Hard) | 86.3% | 82.4% | Inkling 2 Small |
| SWEBench Pro v2 (Full) | 97.7% | 95% | Inkling 2 Small |
| SWEBench Verified | 82.2% | 80% | Inkling 2 Small |
| Search and Tools | |||
| MCP Atlas · Full / Public | 85.4% | 89.2% | Muse Spark Open |
| Knowledge Work | |||
| AA-Briefcase v1.1 | 1,527 | 1,279 | Inkling 2 Small |
| GDPVal-AA v2.1 | 1,630 | 1,401 | Inkling 2 Small |
| JobBench | 70.5% | 61.8% | Inkling 2 Small |
| Enterprise Operations | |||
| AutomationBench v1.0.6 | 45.6% | 49.7% | Muse Spark Open |
| Finance Work | |||
| Finance Agent v2.1 | 67.5% | 60.7% | Inkling 2 Small |
| Excel Modeling Benchmark v1.1 | 69.5% | 62% | Inkling 2 Small |
| Tax Agent Bench | 65.8% | 52.1% | Inkling 2 Small |
| Apex-Accounting v1.1 · pass@1 / mean score | 14.5% | 13.8% | Inkling 2 Small |
| Diligence Bench · Best harness | 58.1% | 59% | Muse Spark Open |
| Legal Work | |||
| Harvey LAB v1.1 | 21.7% | 25% | Muse Spark Open |
| Legal Research Bench | 46.6% | 38.5% | Inkling 2 Small |
| Visual Creation | |||
| DesignArena Website | 1,340 | 1,303 | Inkling 2 Small |
| DesignArena Game Dev | 1,428 | 1,270 | Inkling 2 Small |
| DesignArena 3D Design | 1,401 | 1,261 | Inkling 2 Small |
| Visual Perception | |||
| Humanity’s Sixth Sense | 33.3% | 29.6% | Inkling 2 Small |
| BabyVision · original / with Python | 68.6% | 44.3% | Inkling 2 Small |
| Visual Reasoning | |||
| GDP.pdf | 25.2% | 18.6% | Inkling 2 Small |
| MMMU Pro (Standard 10) | 81.6% | 80.9% | Inkling 2 Small |
| Charxiv RQ · original / with Python | 85.6% | 85.4% | Inkling 2 Small |
| Chartography · original / with Python | 36.7% | 35.7% | Inkling 2 Small |
| MathVision | 92.8% | 87.7% | Inkling 2 Small |
| SAGE v1.1 | 88.6% | 85.8% | Inkling 2 Small |
| STEM Reasoning | |||
| CritPt v2 · Full / Hard | 55.7% | 52.6% | Inkling 2 Small |
| HLE-Diamond | 28.7% | 26.3% | Inkling 2 Small |
| HLE (text only) | 44.1% | 43.9% | Inkling 2 Small |
| GPQA Diamond | 93% | 89% | Inkling 2 Small |
| Apex 2025 | 63.5% | 32.3% | Inkling 2 Small |
| HMMT Feb 2026 | 95.8% | 93.9% | Inkling 2 Small |
| Long Context | |||
| AA-LCR v1.1 | 83.3% | 78.3% | Inkling 2 Small |
| LongBench v2 | 70.6% | 68.7% | Inkling 2 Small |
| Factuality | |||
| AA-Omniscience v1.1 · score | 16.1 | 25.8 | Muse Spark Open |
| AA-Omniscience v1.1 · non-hallucination rate | 73.3% | 72% | Inkling 2 Small |
| Instruction Following | |||
| ComplexConstraints v1.0 | 44.5% | 41.3% | Inkling 2 Small |
| Safety | |||
| DistressBench | 82% | 87.7% | Muse Spark Open |
| FORTRESS (adversarial) | 83.8% | 92.4% | Muse Spark Open |
| FORTRESS (benign) | 79.7% | 84% | Muse Spark Open |
| StrongREJECT (none) | 97.1% | 97.1% | tie |
| Cybersecurity | |||
| PatchEval | 79.6% | 77.8% | Inkling 2 Small |
Reading the comparison honestly
These numbers come from one table published by Thinking Machines, using third-party evaluations where those exist. That makes them comparable to each other, but it does not make them neutral: the vendor chose which benchmarks to run and how to present them, and a benchmark that a model loses has a way of not appearing.
The models are also not the same shape of thing. Muse Spark Open sits in the open weights, larger class tier of the card's own grouping, while Inkling 2 Small is the subject of this site. A win against a smaller model is less meaningful than a win against a larger one, and the table does not adjust for that.