Inkling 2 Small vs Kimi K3
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 65 benchmarks where both models report a score, Inkling 2 Small is higher on 39 and Kimi K3 is higher on 25 , 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 GDPVal-AA v2.1. Its largest deficit is on Code2Video-AA.
Head to head
- Position
- Inkling 2 Small: the subject of this site · Kimi K3: open weights, larger class
- Bechmarks scored
- 65 shared rows, out of 71 total
- Record
- 39 wins · 25 losses · 1 ties
Largest differences
| Benchmark | Inkling 2 Small | Kimi K3 | Gap |
|---|---|---|---|
| GDPVal-AA v2.1 · Knowledge Work | 1,630 | 1,543 | + 87 |
| DesignArena Game Dev · Visual Creation | 1,428 | 1,351 | + 77 |
| PMPP-Hard · Autoresearch | 494.27 | 453.58 | + 40.69 |
| AA-3D · Visual Creation | 1,021 | 982 | + 39 |
| AA-Briefcase v1.1 · Knowledge Work | 1,527 | 1,497 | + 30 |
| EnterpriseBench-CoreCraft · Enterprise Operations | 71.8% | 43.1% | + 28.7 |
| Code2Video-AA · Visual Creation | 1,020 | 1,041 | -21 |
| DesignArena 3D Design · Visual Creation | 1,401 | 1,381 | + 20 |
| FORTRESS (benign) · Safety | 79.7% | 96% | -16.3 |
| JobBench · Knowledge Work | 70.5% | 54.3% | + 16.2 |
Every shared benchmark
| Benchmark | Inkling 2 Small | Kimi K3 | Leader |
|---|---|---|---|
| Third Party Index | |||
| Vals Index (v2.2) | 56.1% | 53% | Inkling 2 Small |
| Coding | |||
| FrontierCode v2.0 | 25.4% | 17.3% | Inkling 2 Small |
| DeepSWE v1.1 | 72.1% | 68.5% | Inkling 2 Small |
| Terminal Bench Science v0.1 | 5.7% | 8.1% | Kimi K3 |
| Terminal Bench v4.0 | 13.1% | 12.6% | Inkling 2 Small |
| Terminal Bench 2.1 · Inkling 2 Small: Pi harness | 85.1% | 88.3% | Kimi K3 |
| Terminal Bench v2.1 (Terminus) | 81.3% | 85% | Kimi K3 |
| Vibe Code Bench v1.1 | 81.6% | 85% | Kimi K3 |
| Sudo L7 | 14.1% | 13.9% | Inkling 2 Small |
| SWE-Marathon v1.1 | 38.8% | 48.1% | Kimi K3 |
| SWEBench Multilingual | 81.2% | 91.4% | Kimi K3 |
| SWEBench Multimodal v2 | 43.1% | 44.2% | Kimi K3 |
| SWEBench Pro v2 (Hard) | 86.3% | 88.2% | Kimi K3 |
| SWEBench Pro v2 (Full) | 97.7% | 97.7% | tie |
| SWEBench Verified | 82.2% | 93.4% | Kimi K3 |
| Autoresearch | |||
| PostTrainBench v1.2 | 31.9% | 31.5% | Inkling 2 Small |
| PMPP-Hard · AUC (task-hours, 0–10h) | 494.27 | 453.58 | Inkling 2 Small |
| BioMysteryBench | 70% | 72.6% | Kimi K3 |
| Search and Tools | |||
| DeepSearchQA | 90.9% | 95% | Kimi K3 |
| MCP Atlas · Full / Public | 85.4% | 82.3% | Inkling 2 Small |
| Knowledge Work | |||
| AA-Briefcase v1.1 | 1,527 | 1,497 | Inkling 2 Small |
| GDPVal-AA v2.1 | 1,630 | 1,543 | Inkling 2 Small |
| JobBench | 70.5% | 54.3% | Inkling 2 Small |
| Enterprise Operations | |||
| AutomationBench v1.0.6 | 45.6% | 58.3% | Kimi K3 |
| EnterpriseOps-Gym | 46.8% | 45.3% | Inkling 2 Small |
| EnterpriseBench-CoreCraft | 71.8% | 43.1% | Inkling 2 Small |
| Finance Work | |||
| Tau 3 Banking v1.0.1 | 40.4% | 33.4% | Inkling 2 Small |
| Finance Agent v2.1 | 67.5% | 60% | Inkling 2 Small |
| Excel Modeling Benchmark v1.1 | 69.5% | 71.3% | Kimi K3 |
| Tax Agent Bench | 65.8% | 68.7% | Kimi K3 |
| DAYJOB (Finance) | 14.3% | 5.5% | Inkling 2 Small |
| Apex-Accounting v1.1 · pass@1 / mean score | 14.5% | 9.5% | Inkling 2 Small |
| Legal Work | |||
| Harvey LAB v1.1 | 21.7% | 12.9% | Inkling 2 Small |
| Legal Research Bench | 46.6% | 46.2% | Inkling 2 Small |
| Visual Creation | |||
| DesignArena Web Apps | 1,296 | 1,283 | Inkling 2 Small |
| DesignArena Website | 1,340 | 1,334 | Inkling 2 Small |
| DesignArena Game Dev | 1,428 | 1,351 | Inkling 2 Small |
| DesignArena 3D Design | 1,401 | 1,381 | Inkling 2 Small |
| Code Arena | 1,653 | 1,656 | Kimi K3 |
| AA-3D | 1,021 | 982 | Inkling 2 Small |
| Code2Video-AA | 1,020 | 1,041 | Kimi K3 |
| Visual Perception | |||
| PerceptionBench | 64.7% | 58.5% | Inkling 2 Small |
| ScreenSpot Pro · original / with Python | 83.8% | 73.8% | Inkling 2 Small |
| Humanity’s Sixth Sense | 33.3% | 25.5% | Inkling 2 Small |
| BabyVision · original / with Python | 68.6% | 53.4% | Inkling 2 Small |
| Visual Reasoning | |||
| GDP.pdf | 25.2% | 22% | Inkling 2 Small |
| MMMU Pro (Standard 10) | 81.6% | 80.5% | Inkling 2 Small |
| Charxiv RQ · original / with Python | 85.6% | 84.8% | Inkling 2 Small |
| Chartography · original / with Python | 36.7% | 26.6% | Inkling 2 Small |
| MathVision | 92.8% | 93.1% | Kimi K3 |
| SAGE v1.1 | 88.6% | 86.8% | Inkling 2 Small |
| STEM Reasoning | |||
| CritPt v2 · Full / Hard | 55.7% | 40.5% | Inkling 2 Small |
| HLE-Diamond | 28.7% | 26.9% | Inkling 2 Small |
| HLE (text only) | 44.1% | 46.9% | Kimi K3 |
| GPQA Diamond | 93% | 93.5% | Kimi K3 |
| Apex 2025 | 63.5% | 65.6% | Kimi K3 |
| HMMT Feb 2026 | 95.8% | 97% | Kimi K3 |
| Long Context | |||
| AA-LCR v1.1 | 83.3% | 88.7% | Kimi K3 |
| LongBench v2 | 70.6% | 69% | Inkling 2 Small |
| Instruction Following | |||
| ComplexConstraints v1.0 | 44.5% | 37.9% | Inkling 2 Small |
| Safety | |||
| DistressBench | 82% | 73.5% | Inkling 2 Small |
| FORTRESS (adversarial) | 83.8% | 72.8% | Inkling 2 Small |
| FORTRESS (benign) | 79.7% | 96% | Kimi K3 |
| StrongREJECT (none) | 97.1% | 99% | Kimi K3 |
| Cybersecurity | |||
| CyberBench v1.1 | 72.7% | 75.2% | Kimi K3 |
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. Kimi K3 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.