Inkling 2 Small vs Inkling Small
Full profiles: Inkling 2 Small (Thinking Machines) and Inkling Small (Thinking Machines): every benchmark, price and rank.
Updated · Data from 8 sources
As of 6 Nov 2026, Inkling 2 Small and Inkling Small are neck and neck on quality (57.4 vs 52.1 on the BenchLeader Index, inside the ±6.8 margin), and they cost about the same at $0.475 vs $0.525 per 1M tokens blended. Inkling 2 Small leads in every category we measure both in.
Key facts (as of 6 Nov 2026)
- Inkling 2 Small scores 57.4 on the BenchLeader Index and Inkling Small scores 52.1 (BenchLeader, 6 Nov 2026).
- Inkling 2 Small costs $0.475 per 1M tokens blended (list price, 6 Nov 2026).
- Inkling Small costs $0.525 per 1M tokens blended (list price, 6 Nov 2026).
- Of the 31 benchmarks both models have, Inkling 2 Small leads 30 (BenchLeader, 6 Nov 2026).
Which to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| Coding | Inkling 2 Small | 56 vs 46 category score; FrontierCode 2.0 (AA) 25.4% vs 4.8% | High |
| Agents & tool use | Inkling 2 Small | 67 vs 46 category score; MCP Atlas 85.4% vs 79.2% | High |
| Reasoning | Inkling 2 Small | 53 vs 44 category score; HLE-Diamond (AA) 28.7% vs 12.1% | High |
| Knowledge | Either | 55 vs 54 category score; AA-Omniscience v1.1 16.1 vs -9.5 | Either |
| Long documents | Inkling 2 Small | 67 vs 63 category score; AA-LCR v1.1 83.3% vs 75.7% | Medium |
| Images and other media | Inkling 2 Small | 64 vs 56 category score; MMMU-Pro 81.6% vs 74.0% | High |
| Chat quality (human preference) | – | Only Inkling Small has an Arena rating (58 category score) | Not comparable |
| Lowest cost | Either | $0.475 vs $0.525 per 1M blended; agentic coding session $0.022 vs $0.023 | – |
| Fastest first answer | Inkling Small | 19.6 s (Inkling Small) vs 36.2 s (Inkling 2 Small) to the first answer token | – |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- Inkling 2 Small and Inkling Small are neck and neck on quality (57.4 vs 52.1, within the ±6.8 margin).
- Inkling 2 Small is stronger in agents & tools, coding, knowledge, long context, multimodal, reasoning.
- Inkling Small is stronger in human preference, maths.
- They cost about the same ($0.475 per 1M blended).
At a glance
| Spec | Inkling 2 Small | Inkling Small |
|---|---|---|
| Developer | Thinking Machines | Thinking Machines |
| Released | 5 Nov 2026 | 15 Jul 2026 |
| Context window (tokens) | 1,024,000 | 65,536 |
| Compared at | Inkling 2 Small | Inkling Small |
| Open weights | No | Yes |
| Price in / out / cached input per 1M | $0.300 / $1.00 / – | $0.300 / $1.20 / $0.116 |
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | Inkling 2 Small | Inkling Small | Leader |
|---|---|---|---|
| Reasoning | 52.6 | 43.8 | Inkling 2 Small |
| Coding | 56.3 | 45.8 | Inkling 2 Small |
| Agents & tools | 66.9 | 45.6 | Inkling 2 Small |
| Maths | – | 52.9 | not comparable |
| Knowledge | 55.0 | 54.3 | level |
| Human preference | – | 57.9 | not comparable |
| Multimodal | 64.1 | 56.4 | Inkling 2 Small |
| Long context | 66.5 | 62.6 | Inkling 2 Small |
On one chart
Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.
Not drawn, measured for fewer than two of these models: Maths, Instruction following, Human preference.
Where sources disagree
Across the 31 benchmarks both have, Inkling 2 Small leads 30, Inkling Small leads 0 and 1 is a tie. Every overall index that rates both puts Inkling 2 Small ahead.
| Overall index | Inkling 2 Small | Inkling Small | Leader |
|---|---|---|---|
| BenchLeader Index | 57.4 | 52.1 | neck and neck |
| AA Intelligence Index v5.0 | 30.6 | 12.9 | Inkling 2 Small |
| Vals Index | 56.1 | 27.6 | Inkling 2 Small |
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.
| Measure | Inkling 2 Small | Inkling Small |
|---|---|---|
| Blended price per 1M tokens | $0.475 | $0.525 |
| Input / output per 1M | $0.300 / $1.00 | $0.300 / $1.20 |
| Cached input per 1M | – | $0.116 |
| Price per task: chat reply | $0.0004 | $0.0005 |
| Price per task: summarize a 30-page report | $0.0042 | $0.0043 |
| Price per task: code edit | $0.0033 | $0.0036 |
| Price per task: agentic coding session | $0.022 | $0.023 |
| Price per task: structured extraction | $0.0008 | $0.0008 |
| Agentic coding session, input mostly cached | – | $0.015 |
| Cost per benchmark task (AA Index) | $1.18 | $0.203 |
| Output tokens per task (AA Index) | 150k | 37k |
| Of which reasoning | 88k | 25k |
| Time per task (AA Index) | 21 min | 5 min |
| Output speed | 118 tok/s | 129 tok/s |
| Time to first answer | 36.2 s | 19.6 s |
| Time to a complete answer | 105.2 s | 52.6 s |
Per token is not per task: an average task of Artificial Analysis' Intelligence Index cost $1.18 on Inkling 2 Small and $0.203 on Inkling Small, and Inkling 2 Small wrote 4.1× as many output tokens for the same work, which a per-token price hides.
Each model is compared at its best-ranked reasoning-effort setting. Inkling 2 Small scores 57.4 and takes 36 s to its first answer token. Inkling Small scores 52.1 and takes 20 s to its first answer token; Inkling Small (xhigh) scores 47.5 in 20 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
96 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | Inkling 2 Small | Inkling Small |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 57.4 | 52.1 |
| Agents & tools score | BenchLeader | 66.9 | 45.6 |
| Coding score | BenchLeader | 56.3 | 45.8 |
| Knowledge score | BenchLeader | 55.0 | 54.3 |
| Long context score | BenchLeader | 66.5 | 62.6 |
| Multimodal score | BenchLeader | 64.1 | 56.4 |
| Reasoning score | BenchLeader | 52.6 | 43.8 |
| Human preference score | BenchLeader | – | 57.9 |
| Maths score | BenchLeader | – | 52.9 |
| Blended price $/M | BenchLeader | $0.475 | $0.525 |
| Output speed | BenchLeader | 118 tok/s | 129 tok/s |
| Time to first answer | BenchLeader | 36.2 s | 19.6 s |
| Context window | BenchLeader | 1,024,000 tokens | 65,536 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | – | 88.5%† |
| Arena Hard Prompts | Arena (LMArena) | – | 1427 |
| GPQA Diamond (AA) | Artificial Analysis | – | 89.5% |
| Humanity's Last Exam (AA) | Artificial Analysis | 44.1% | 33.3% |
| HLE-Diamond (AA) | Artificial Analysis | 28.7% | 12.1% |
| GPQA Diamond (Vals) | Vals AI | – | 83.6% |
| ARC-AGI-1 | ARC Prize | – | 84.0%† |
| ARC-AGI-2 | ARC Prize | – | 40.1%† |
| CritPt | Artificial Analysis | – | 8.3% |
| CritPt v2 hard subset (AA) | Artificial Analysis | 35.0% | 18.5% |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | – | 6.0%† |
| Coding | |||
| SciCode | SciCode | – | 49.6% |
| Arena Coding | Arena (LMArena) | – | 1472 |
| Arena WebDev | Arena (LMArena) | – | 1409 |
| SciCode (AA) | Artificial Analysis | 52.3% | 49.6% |
| LiveCodeBench | Vals AI | – | 85.9% |
| IOI | Vals AI | – | 9.3% |
| SWE-bench (Vals) | Vals AI | – | 82.2% |
| Code Migration | Vals AI | – | 13.7% |
| Code Migration v1.1 | Vals AI | – | 10.4% |
| ProgramBench | Vals AI | 0.5% | 0.5% |
| SWE-bench Multimodal | Vals AI | 43.1% | – |
| Vibe Code Bench v1.1 | Vals AI | 81.5% | 19.1% |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 13.1% | 1.0% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 81.3% | 55.1% |
Show all 96 benchmarks
| Metric | Publisher | Inkling 2 Small | Inkling Small |
|---|---|---|---|
| Coding | |||
| GameBench 2 Overall | DuelLab GameBench 2 | – | 29.8 |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | – | 28.1† |
| FrontierCode 1.1 (AA) | Artificial Analysis | – | 0.6% |
| FrontierCode 2.0 (AA) | Artificial Analysis | 25.4% | 4.8% |
| Agents & tools | |||
| Arena Agent | Arena (LMArena) | – | -17 |
| Terminal-Bench 2.1 (Vals) | Vals AI | – | 55.1% |
| MCP Atlas | Scale AI SEAL | 85.4% | 79.2% |
| GDPval-AA v2.1, normalised score | Artificial Analysis | – | 31.2% |
| GDPval-AA v2.1 | Artificial Analysis | 1630 | 1133 |
| τ³-Banking (AA) | Artificial Analysis | 40.4% | 18.8% |
| Analyst Agent (AA) | Artificial Analysis | 51.3% | 27.5% |
| CyberBench | Vals AI | 72.7% | 64.1% |
| Finance Agent v2 | Vals AI | – | 41.3% |
| Finance Agent v2.1 | Vals AI | 67.5% | 44.8% |
| Harvey's Legal Agent Benchmark v1.1 | Vals AI | 21.7% | 2.9% |
| Harvey's Legal Agent Benchmark v1 | Vals AI | – | 1.7% |
| Legal Research Bench | Vals AI | 46.6% | 25.5% |
| SkillsBench | Vals AI | – | 33.6% |
| SREBench | Vals AI | 1.9% | – |
| Tax Agent Bench | Vals AI | 38.9% | 15.4% |
| Terminal-Bench 4.0 (Vals) | Vals AI | 12.6% | 0.0% |
| Terminal-Bench Science | Vals AI | 2.9% | – |
| AutomationBench | Artificial Analysis | 45.6% | 4.8% |
| GDP.pdf (AA) | Artificial Analysis | 25.2% | 11.2% |
| Harvey LAB-AA v1.1 | Harvey | – | 0.0% |
| EnterpriseOps-Gym | Artificial Analysis | 46.8% | 42.7% |
| AA-Briefcase v1.1 | Artificial Analysis | 1527 | 913 |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 5.7% | 0.0% |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | – | 46.3%† |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | – | 17.1%† |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | – | 90.0%† |
| ProofBench | Vals AI | – | 6.0% |
| Arena Maths | Arena (LMArena) | – | 1451 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | – | 19.1%† |
| AA-Omniscience v1.1 | Artificial Analysis | 16.1 | -9.5 |
| MMLU-Pro | Vals AI | – | 85.6% |
| LegalBench | Vals AI | – | 83.0% |
| CorpFin | Vals AI | – | 69.6% |
| TaxEval | Vals AI | – | 75.5% |
| BioMysteryBench | Vals AI | 70.0% | – |
| Excel Modeling Benchmark | Vals AI | – | 31.8% |
| Excel Modeling Benchmark v1.1 | Vals AI | 69.5% | 40.8% |
| MedCode | Vals AI | – | 37.9% |
| MedScribe | Vals AI | – | 84.1% |
| Public Benefits Bench v1.1 | Vals AI | – | 59.4% |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 33.8% | 31.8% |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 73.3% | 39.5% |
| Instruction following | |||
| Arena Instruction Following | Arena (LMArena) | – | 1396 |
| Human preference | |||
| Arena Text | Arena (LMArena) | – | 1406 |
| Arena Creative Writing | Arena (LMArena) | – | 1318 |
| Arena Multi-turn | Arena (LMArena) | – | 1414 |
| Arena Longer Queries | Arena (LMArena) | – | 1402 |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | – | 1237 |
| MMMU-Pro | Artificial Analysis | 81.6% | 74.0% |
| MortgageTax | Vals AI | – | 62.3% |
| SAGE | Vals AI | 88.6% | – |
| AudioMC | Scale AI SEAL | – | 54.9% |
| AudioMC (text output) | Scale AI SEAL | – | 54.9% |
| Code2Video-AA | Artificial Analysis | 1020 | – |
| AA-3D | Artificial Analysis | 1021 | – |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 83.3% | 75.7% |
| MLCR | Artificial Analysis | – | 7.8% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | – | 150.8 |
| AA Intelligence Index v5.0 | Artificial Analysis | 30.6 | 12.9 |
| AA Intelligence Index v4.3.2 | Artificial Analysis | – | 25.7 |
| AA Progress Index v1 | Artificial Analysis | – | 282.7 |
| Vals Index | Vals AI | 56.1 | 27.6 |
| Vals Finance Index | Vals AI | 61.5% | 36.3% |
| Vals Multimodal Index | Vals AI | – | 50.0% |
| Vals Router Index | Vals AI | – | 27.6% |
| Metadata | |||
| AA Openness Index | Artificial Analysis | – | 33.3 |
Data as of 2026-11-06. Best configuration of each model; every score links to its source on the model pages.
DuelLab GameBench 2, release gb2-20261103-09b2218cbd85 (https://benchmarks.duellab.org/releases/gb2-20261103-09b2218cbd85/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
What changed
- Inkling 2 Small: first results.
- Inkling Small: 1 score revision.
- Inkling Small: 1 score revision.
- Inkling Small: BenchLeader Index 54.9 → 51.9.
- Inkling Small: 5 new results (AA Intelligence Index v5.0, AA Progress Index v1, HLE-Diamond (AA)…); 3 score revisions; Inkling Small cost per task moved from $0.089/M to $0.20/M (127% up).
- Inkling Small: 1 new result (AA-Briefcase v1.1).
- Inkling Small: 6 new results (Code Migration v1.1, Excel Modeling Benchmark v1.1, Finance Agent v2.1…); Inkling Small cached input price moved from $0.10/M to $0.12/M (16% up).
- Inkling Small: 2 new results (MLCR, FrontierCode 1.1 (AA)).
- Inkling Small: 3 score revisions.
- Inkling Small: 5 new results (AutomationBench, GDP.pdf (AA), Harvey LAB-AA v1.1…).
Related comparisons
- Inkling 2 Small vs DeepSeek V4 Pro
- Inkling 2 Small vs MiMo-V2.6-Pro
- Inkling 2 Small vs DeepSeek V4 Flash
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
Inkling 2 Small vs Inkling Small: questions
- Is Inkling 2 Small better than Inkling Small?
- Inkling 2 Small and Inkling Small are neck and neck on quality (57.4 vs 52.1, within the ±6.8 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 6 Nov 2026, so check the category scores for your use.
- Inkling 2 Small vs Inkling Small: which is cheaper, and will that change?
- Inkling 2 Small is cheaper today: $0.475 against $0.525 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is Inkling 2 Small or Inkling Small better for coding?
- Inkling 2 Small scores higher in coding (56 vs 46 on the category index, where 50 is average). On FrontierCode 2.0 (AA), the top shared coding benchmark, it is 25.4% for Inkling 2 Small against 4.8% for Inkling Small.
- Which is faster, Inkling 2 Small or Inkling Small?
- Inkling Small streams faster (129 vs 118 output tokens per second, Artificial Analysis 7-day median).
- Is Inkling 2 Small or Inkling Small better for agents and tool use?
- Inkling 2 Small scores higher in agentic tasks (67 vs 46 on the category index, where 50 is average). On MCP Atlas, the top shared agents & tools benchmark, it is 85.4% for Inkling 2 Small against 79.2% for Inkling Small.
More questions (4)
- Which is better for long documents?
- Inkling 2 Small scores higher on long-context benchmarks (67 vs 63). Inkling 2 Small accepts more context: 1,024,000 against 65,536 tokens for Inkling Small.
- Which has lower latency?
- Inkling Small answers first: 19.6 s against 36.2 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
- Do independent benchmarks agree on Inkling 2 Small vs Inkling Small?
- Across the 31 benchmarks both have, Inkling 2 Small leads 30, Inkling Small leads 0 and 1 is a tie. Every overall index that rates both puts Inkling 2 Small ahead.
- When was Inkling 2 Small released?
- Inkling 2 Small was released on 5 Nov 2026.