Thinking Machines Lab: inkling-small
2 benchmarks
Leaderboard position
#148
Overall rank (of 1121 models)
2059
Elo rating
2
Benchmarks
Top 13.2%
Percentile
In Thinking Machines Lab (of 3 models)
Above: Anthropic: claude-4.5-sonnet #147
Below: xtools-at: deepseek-v4-pro-qwen3.5-9b-uncensored-heretic #149
Benchmark highlights
Best benchmark
Artificial Analysis Intelligence Index v4.3.2
26
Worst benchmark
Artifical Analysis Intelligence Index v5.0
13
Nearest competitor from another company
The model from another company with the most similar benchmark scores.
appeared as nearest in 1 of 2 benchmarks
1/2
50.0%
| Model | Votes | % of benchmarks |
|---|---|---|
| Qwen: qwen3.7-plus | 1 | 50.0% |
| NVIDIA Corporation: nvidia-nemotron-3.5-super-vl | 1 | 50.0% |
Benchmarks
Predicted scores
Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.
| Benchmark | Predicted | Confidence | Based on |
|---|---|---|---|
| MATH-500 | ~99.2 | 50% | — |
| AIME 2025 | ~100 | 50% | — |
| AIME 2026 | ~96.79 | 50% | — |
| HMMT Feb 2026 | ~88.39 | 50% | — |
| GPQA Diamond | ~84.58 | 50% | — |
| MMLU-Pro | ~90.53 | 50% | — |
| Humanity's Last Exam (no tools) | ~33.07 | 50% | — |
| IFEval | ~96.75 | 50% | — |
| IFBench | ~73.28 | 50% | — |
| HumanEval | ~95.9 | 50% | — |
| MBPP | ~93.7 | 50% | — |
| Terminal-Bench v2.1 | ~63.26 | 50% | — |
| MMLU | ~92.5 | 50% | — |
| SWE-bench Verified | ~76.98 | 50% | — |
| SWE-bench Pro | ~53.38 | 50% | — |
| SciCode | ~45.74 | 50% | — |
| AA-LCR | ~66.63 | 50% | — |
| Artificial Analysis Intelligence Index v5.0.1 | ~16.87 | 50% | — |
| Humanity's Last Exam (with tools) | ~44.48 | 50% | — |
| GSM8K | ~99.6 | 50% | — |
| -(CoderPPL) | ~-1.69 | 50% | — |
| LiveCodeBench v6 | ~83.45 | 50% | — |
| MMMU-Pro | ~75.57 | 50% | — |
| MATH | ~97.9 | 50% | — |
| BBH | ~88.4 | 50% | — |