Thinking Machines Lab
availableShows if the model has enough results for an index.Inkling-Small
Inkling-Small is a hybrid model from Thinking Machines Lab in the Inkling family. 41 benchmarks count toward its score, in 7 categories.
IndexOverall score out of 100.56.9 ±3.0
CoverageShare of the index weight with results.95%
SpeedOutput tokens per second.98/s
Input / 1MUS dollars per 1M input tokens.$0.45
Output / 1MUS dollars per 1M output tokens.$1.2
ContextMaximum tokens in one request.1.05M
EloLMArena rating and rank.1412 (#126)
The index is a score out of 100. The ± range shows how much it can change.
18,844 votes. Elo shows what people prefer. It does not change the score.
CapabilitiesScore per category, out of 100.
Out of 100Results
41 counted| BenchmarkThe test name. | CategoryThe capability that the test measures. | ResultThe score from the publisher. | IndexThis result as a score out of 100. | RunThe settings of the run. | DateDate of the result. | Published byThe source of the result. |
|---|---|---|---|---|---|---|
| AIME 2026 | Math | 95.5% | 54.6 | — | — | Qwen |
| Harvard-MIT Mathematics Tournament February 2026 | Math | 90.2% | 57.7 | — | — | Qwen |
| OTIS Mock AIME 2024-2025 | Math | 90.0% | 61.9 | xhigh effort | — | Epoch AI |
| Graduate-Level Google-Proof Q&A | Knowledge | 89.5% | 60.9 | — | — | David Rein et al. |
| GPQA Diamond | Knowledge | 89.5% | 60.9 | — | — | David Rein et al. |
| Artificial Analysis GPQA Diamond | Knowledge | 89.5% | 60.3 | — | — | Artificial Analysis |
| GPQA diamond | Knowledge | 88.5% | 60.0 | xhigh effort | — | Epoch AI |
| LiveCodeBench | Coding | 85.9% | 62.4 | — | 1 Sept 2026 | Vals AI |
| MMLU Pro | Knowledge | 85.6% | 55.4 | — | 1 Sept 2026 | Vals AI |
| ARC-AGI-1 (semi-private) | Reasoning | 84.0% | 66.9 | xhigh effort | — | ARC Prize Foundation |
| GPQA Diamond | Knowledge | 83.6% | 55.4 | — | 1 Sept 2026 | Vals AI |
| Instruction Following Benchmark | Instruction | 82.2% | 61.7 | — | — | Benchmark authors |
| SWE-bench | Coding | 82.2% | 63.5 | — | 1 Sept 2026 | Vals AI |
| CharXiv Reasoning | Multimodal | 81.3% | 58.7 | — | — | CharXiv authors |
| Software Engineering Benchmark Verified | Coding | 80.2% | 61.9 | — | — | Carlos E. Jimenez et al. |
| MCP Atlas | Agentic | 79.6% | 68.1 | — | — | OpenAI |
| BrowseComp | Agentic | 77.4% | 64.8 | — | — | OpenAI |
| CharXiv Reasoning without tools | Multimodal | 77.4% | — | — | — | CharXiv authors |
| Artificial Analysis Long Context Reasoning | Reasoning | 75.7% | 60.6 | — | — | Artificial Analysis |
| Massive Multi-discipline Multimodal Understanding Pro | Multimodal | 74.0% | 47.8 | — | — | MMMU-Pro authors |
| Artificial Analysis MMMU-Pro | Multimodal | 74.0% | 57.5 | — | — | Artificial Analysis |
| SWE-bench Pro | Coding | 55.9% | 58.1 | — | — | Xiang Deng et al. |
| Terminal-Bench 2.1 | Agentic | 55.1% | 56.5 | — | 21 Sept 2026 | Vals AI |
| Toolathlon-Verified | Agentic | 54.4% | 55.6 | — | — | Moonshot AI |
| Artificial Analysis Coding Index | Coding | 53.0% | 56.3 | — | — | Artificial Analysis |
| Artificial Analysis SciCode | Coding | 49.7% | 61.8 | — | — | Artificial Analysis |
| Scientific Code Benchmark | Coding | 48.7% | 59.2 | — | — | Benchmark authors |
| Humanity's Last Exam | Knowledge | 47.8% | 69.3 | — | — | Center for AI Safety et al. |
| FrontierMath-Tiers-1-3-v2-Private | Math | 46.3% | 57.5 | xhigh effort | — | Epoch AI |
| ARC-AGI-2 (semi-private) | Reasoning | 40.1% | 60.0 | xhigh effort | — | ARC Prize Foundation |
| SkillsBench | Coding | 33.6% | 51.5 | OpenHands | 11 Sept 2026 | Vals AI |
| Artificial Analysis Humanity's Last Exam | Knowledge | 33.3% | 60.7 | — | — | Artificial Analysis |
| Artificial Analysis Omniscience Accuracy | Knowledge | 33.2% | 54.9 | — | — | Artificial Analysis |
| Humanity's Last Exam without tools | Knowledge | 31.6% | 55.6 | — | — | OpenAI |
| Artificial Analysis Intelligence Index | Knowledge | 27.8% | 57.1 | — | — | Artificial Analysis |
| Artificial Analysis Agentic Index | Agentic | 24.9% | 57.5 | — | — | Artificial Analysis |
| SimpleQA Verified | Knowledge | 19.1% | 39.2 | xhigh effort | — | Epoch AI |
| Vibe Code Bench v1.1 | Coding | 19.1% | 50.1 | OpenHands | 21 Sept 2026 | Vals AI |
| Chess Puzzles | Reasoning | 18.0% | 46.8 | xhigh effort | — | Epoch AI |
| FrontierMath-Tier-4-v2-Private | Math | 17.1% | 56.1 | xhigh effort | — | Epoch AI |
| Code Migration | Coding | 13.7% | 53.7 | — | 21 Sept 2026 | Vals AI |
| IOI | Coding | 9.3% | 49.7 | — | 21 Sept 2026 | Vals AI |
| Critical Physics Tasks | Reasoning | 8.3% | 55.8 | — | — | Artificial Analysis |
| Agent Arena command recovery | Agentic | 6.9 | 75.1 | — | 15 Sept 2026 | LMArena |
| ProofBench v1.1 | Math | 6.0% | 51.8 | — | 21 Sept 2026 | Vals AI |
| Mystery Game Puzzles | Reasoning | 6.0% | 40.3 | xhigh effort | — | Epoch AI |
| Terminal-Bench 4.0 | Agentic | 1.5% | 60.5 | — | 21 Sept 2026 | Vals AI |
| ProgramBench | Coding | 0.5% | — | — | 21 Sept 2026 | Vals AI |
| Agent Arena steerability | Agentic | -11.8 | 54.0 | — | 15 Sept 2026 | LMArena |
| Agent Arena task outcome | Agentic | -20.4 | 44.4 | — | 15 Sept 2026 | LMArena |
41 benchmarks count, from 48 of 50 results. A grey row does not count. Too few models took that benchmark.