Inkling Small
Inkling Small is a Thinking Machines open-weights reasoning model, released 15 Jul 2026. Its best configuration ranks #240 of 785 on the BenchLeader Index at 54.8 ±5.2, in the upper half. The ± is the point: 301 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in long context (63) and lowest in agents & tools (46). At $0.525 per million tokens blended it is mid-priced. Output speed of 224 tokens per second puts it in the fastest quarter, with a first answer in 10.5 s. It has been measured at 7 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 22 Oct 2026.
- Blended price
- $0.525/M
- $0.300 in · $1.20 out
- Output speed
- 224 tok/s
- measured by Artificial Analysis
- First answer
- 11 s
- first token 0.42 s
- Context
- 524k
- Full answer
- 13 s
- median, reasoning included
- Cost per run
- $0.089
- one full Intelligence Index run
- Released
- 15 Jul 2026
- Overall index55
- Reasoning56
- Coding50
- Agents & tools46
- Maths53
- Knowledge59
- Human preference58
- Multimodal57
- Long context63
- Composite58
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | – | – | 224 tok/s | 11 s | $0.0005 | Reasoning 29 |
| minimal | – | – | 224 tok/s | 11 s | $0.0005 | Reasoning 37 |
| low | – | – | 224 tok/s | 11 s | $0.0005 | Reasoning 38 |
| medium | – | – | 224 tok/s | 11 s | $0.0005 | Reasoning 43 |
| high | – | – | 224 tok/s | 11 s | $0.0005 | Reasoning 51 |
| xhigh | 47.6 | #452 | 224 tok/s | 11 s | $0.0005 | Knowledge 31 · Maths 50 · Reasoning 58 |
| not statedbest | 54.8 | #240 | 224 tok/s | 11 s | $0.0005 | Agents & tools 46 · Coding 50 · Composite 58 · Human preference 58 · Knowledge 59 · Long context 63 · Maths 53 · Multimodal 57 · Reasoning 56 |
How its index has moved
8 Sept 2026 to 21 Oct 2026- not stated
- xhigh
The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | minimal | low | medium | high | xhigh | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| GPQA Diamond | – | – | – | – | – | 88.5%#59 | – | Epoch AI Benchmarking Hub | |
| Arena Hard Promptseffort not stated | – | – | – | – | – | – | 1426#141 | Arena (LMArena) | |
| GPQA Diamond (AA)not in indexeffort not stated | – | – | – | – | – | – | 89.5%#74 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in indexeffort not stated | – | – | – | – | – | – | 33.3%#137 | Artificial Analysis | |
| GPQA Diamond (Vals)not in indexeffort not stated | – | – | – | – | – | – | 83.6%#55 | Vals AI | |
| ARC-AGI-1 | 5.5%#243 | 48.5%#181 | 52.5%#177 | 67.0%#149 | 78.0%#124 | 84.0%#114 | – | ARC Prize | |
| ARC-AGI-2 | 0.0%#242 | 4.0%#185 | 4.9%#177 | 13.6%#151 | 33.1%#126 | 40.1%#119 | – | ARC Prize | |
| CritPteffort not stated | – | – | – | – | – | – | 8.3%#137 | Artificial Analysis | |
| Mystery Game Puzzlesnot in index | – | – | – | – | – | 6.0%#130 | – | Epoch AI Benchmarking Hub |
Coding
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| SciCode | 49.6%#113 | SciCode | |
| Arena Coding | 1471#118 | Arena (LMArena) | |
| Arena WebDev | 1409#85 | Arena (LMArena) | |
| SciCode (AA)not in index | 49.6%#114 | Artificial Analysis | |
| LiveCodeBench | 85.9%#25 | Vals AI | |
| IOI | 9.3%#42 | Vals AI | |
| SWE-bench (Vals)not in index | 82.2%#20 | Vals AI | |
| Code Migrationnot in index | 13.7%#57 | Vals AI | |
| Code Migration v1.1not in index | 10.4%#26 | Vals AI | |
| ProgramBenchnot in index | 0.5%#18 | Vals AI | |
| Vibe Code Bench v1.1not in index | 19.1%#86 | Vals AI | |
| Terminal-Bench 4.0 (AA)not in index | 1.0%#128 | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 55.1%#109 | Artificial Analysis | |
| FrontierCode 1.1 (AA)not in index | 0.6%#68 | Artificial Analysis |
Agents & tools
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Agent | -15.8#49 | Arena (LMArena) | |
| Terminal-Bench 2.1 (Vals) | 55.1%#48 | Vals AI | |
| MCP Atlas | 79.2%#12 | Scale AI SEAL | |
| GDPval-AA v2.1not in index | 31.2%#77 | Artificial Analysis | |
| τ³-Banking (AA) | 18.8%#102 | Artificial Analysis | |
| Analyst Agent (AA)not in index | 27.5%#31 | Artificial Analysis | |
| CyberBenchnot in index | 64.1%#32 | Vals AI | |
| Finance Agent v2not in index | 41.3%#60 | Vals AI | |
| Finance Agent v2.1not in index | 44.8%#28 | Vals AI | |
| Harvey's Legal Agent Benchmark v1.1not in index | 2.9%#25 | Vals AI | |
| Harvey's Legal Agent Benchmark v1not in index | 1.7%#48 | Vals AI | |
| Legal Research Benchnot in index | 25.5%#55 | Vals AI | |
| SkillsBenchnot in index | 33.6%#31 | Vals AI | |
| Tax Agent Benchnot in index | 15.4%#52 | Vals AI | |
| Terminal-Bench 4.0 (Vals)not in index | 0.0%#50 | Vals AI | |
| AutomationBenchnot in index | 4.8%#167 | Artificial Analysis | |
| GDP.pdf (AA)not in index | 11.2%#120 | Artificial Analysis | |
| Harvey LAB-AA v1.1not in index | 0.0%#23 | Harvey | |
| EnterpriseOps-Gymnot in index | 42.7%#22 | Artificial Analysis | |
| Terminal-Bench-Science 0.1 (AA)not in index | 0.0%#71 | Artificial Analysis |
Maths
| Benchmark | xhigh | not stated | Source | Trend |
|---|---|---|---|---|
| FrontierMath Tiers 1–3 | 46.3%#56 | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | 17.1%#56 | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | 90.0%#74 | – | Epoch AI Benchmarking Hub | |
| ProofBencheffort not stated | – | 6.0%#69 | Vals AI | |
| Arena Mathseffort not stated | – | 1452#73 | Arena (LMArena) |
Knowledge
| Benchmark | xhigh | not stated | Source | Trend |
|---|---|---|---|---|
| SimpleQA Verified | 19.1%#76 | – | Epoch AI Benchmarking Hub | |
| AA-Omniscienceeffort not stated | – | -8.9#183 | Artificial Analysis | |
| MMLU-Proeffort not stated | – | 85.6%#56 | Vals AI | |
| LegalBencheffort not stated | – | 83.0%#61 | Vals AI | |
| CorpFineffort not stated | – | 69.6%#6 | Vals AI | |
| TaxEvaleffort not stated | – | 75.5%#18 | Vals AI | |
| Excel Modeling Benchmarknot in indexeffort not stated | – | 31.8%#61 | Vals AI | |
| Excel Modeling Benchmark v1.1not in indexeffort not stated | – | 40.8%#30 | Vals AI | |
| MedCodenot in indexeffort not stated | – | 37.9%#72 | Vals AI | |
| MedScribenot in indexeffort not stated | – | 84.1%#37 | Vals AI | |
| Public Benefits Bench v1.1not in indexeffort not stated | – | 59.4%#33 | Vals AI | |
| AA-Omniscience: accuracynot in indexeffort not stated | – | 33.2%#165 | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in indexeffort not stated | – | 37.0%#180 | Artificial Analysis |
Instruction following
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Instruction Followingnot in index | 1394#147 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1405#145 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1316#198 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1414#135 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1400#160 | Arena (LMArena) |
Multimodal
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Vision | 1237#74 | Arena (LMArena) | |
| MMMU-Pro | 74.0%#116 | Artificial Analysis | |
| MortgageTaxnot in index | 62.3%#53 | Vals AI | |
| SAGEnot in index | 32.7%#72 | Vals AI | |
| AudioMCnot in index | 54.9%#4 | Scale AI SEAL | |
| AudioMC (text output)not in index | 54.9%#4 | Scale AI SEAL |
Long context
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA-LCR | 75.7%#154 | Artificial Analysis | |
| MLCRnot in index | 7.8%#128 | Artificial Analysis |
Composite
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Epoch Capabilities Indexnot in index | 150.3#56 | Epoch AI Benchmarking Hub | |
| AA Intelligence Index v4.3.2 | 25.7#161 | Artificial Analysis | |
| Vals Indexnot in index | 27.6#24 | Vals AI | |
| Vals Finance Indexnot in index | 36.3%#28 | Vals AI | |
| Vals Multimodal Indexnot in index | 50.0%#25 | Vals AI | |
| Vals Router Indexnot in index | 27.6%#20 | Vals AI |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| DeepInfra | 162 tok/s | 0.47 s | $0.450 | $1.20 | 524k | fp8 |
Price history
The price above, per 1M tokens, every day it changed.
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.116 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0005 | $0.0004 | 11.9 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0043 | $0.0027 | 13.2 s |
| Code edit | 6,000 / 1,500 | $0.0036 | $0.0028 | 17.2 s |
| Agentic coding session | 60,000 / 4,000 | $0.023 | $0.015 | 28.4 s |
| Structured extraction | 2,000 / 200 | $0.0008 | $0.0006 | 11.4 s |
See also
Questions people ask
- What are Inkling Small's benchmark scores?
- As of 24 Oct 2026, Inkling Small scores 54.8 ±5.2 on the BenchLeader Index, strongest in long context (63) and weakest in agents & tools (46). That rests on 23 independent benchmark results, and every score on this page links to its publisher.
- How does Inkling Small rank?
- Inkling Small is #240 of 785 ranked configurations on 24 Oct 2026. 301 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Inkling Small cost per 1M tokens?
- As of 24 Oct 2026, Inkling Small costs $0.300 per 1M input tokens and $1.20 per 1M output tokens, $0.525 blended (three input tokens per output token). The list price last changed on 6 Oct 2026.
- When was Inkling Small released?
- Thinking Machines released Inkling Small on 15 Jul 2026. BenchLeader has results for it at 7 reasoning-effort settings.
Data as of 24 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “Inkling Small: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/inkling-small.