Current model · Thinking Machines Lab
Inkling 2 Small benchmarks, pricing and specs
Inkling 2 Small is a Thinking Machines Lab model released on November 5, 2026 that takes text and image input and produces text. It costs $0.30 per 1M input tokens and $1 per 1M output tokens and accepts up to 262,144 tokens of context.
- Per 1M input tokens
- $0.30Listed today
- Per 1M output tokens
- $1Listed today
- Context window
- 262K#57 of 67 current models
Data verified November 6, 2026 · Dataset 2026-11-06
Specs
Key facts
- Provider
- Thinking Machines Lab
- Model family
- Inkling
- Release date
- November 5, 2026
- Status
- Current
- Access
- API only
- API model id
- thinkingmachines/Inkling-2-Small
- Context window
- 262,144 tokens
- Max output
- Not reported
- Input
- text, image
- Output
- text
- Task families
- text, multimodal
- Declared capabilities
- reasoning, coding, agentic coding, image understanding
Charts
Against every current text model
Where Inkling 2 Small lands among the current models of its task, on every independent benchmark, on price and on context.
Where it stands
10 of 58 list a lower price per 1M input tokens. Each bar is the share of current models it beats.
Price and context
Price per 1M input tokens
#11of 58
$0.30 · cheaper than 75%
Price per 1M output tokens
#10of 58
$1 · cheaper than 84%
Context window
#57of 67
262K tokens · longer than 12%
Benchmarks
Independent benchmark results
Scores come from independent evaluators that ran the model themselves, pinned to a dated snapshot. Rank is among current models in the same snapshot. Thinking Machines Lab's own launch figures are in the next section, because they are not measured the same way.
No independent benchmark has published a result for Inkling 2 Small yet. It is still compared below on price, context and capabilities.
Provider results
Reported by Thinking Machines Lab
Thinking Machines Lab's own results for Inkling 2 Small, beside the rivals it chose, as its table lists them. Labs pick the benchmarks and run them their own way, so compare within one table only. Internal benchmarks and other evaluators' indexes are left out.
Inkling 2 Small model card, November 5, 2026Thinking Machines Lab · 2026-11-06
| Benchmark | Inkling 2 Small | GLM-5.3-Flash | MiMo-V2.6-Flash | MiniMax M3.1 Flash | DeepSeek-V4.1-Flash | Muse Spark Open | GLM-5.3 | MiMo-V2.6-Pro | Mistral Large 4 | Kimi K3 | GPT-6 Luna | Gemini 3.8 Flash | Claude Haiku 5.5 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DeepSWE v1.1 | 72.1% | 63.4% | 67.9% | 73.7% | 77.0% | 60.0% | 69.0% | 71.9% | 61.7% | 68.5% | 66.6% | 73.8% | 59.9% |
| Vibe Code Bench v1.1 | 81.6% | 30.8% | 79.0% | 73.9% | 87.7% | 76.2% | 78.1% | 85.2% | 81.6% | 85.0% | 81.7% | 78.7% | 90.4% |
| Sudo L7 | 14.1% | Not listed | Not listed | Not listed | Not listed | Not listed | 15.0% | Not listed | 13.9% | 13.9% | Not listed | 10.6% | Not listed |
| SWE-Marathon v1.1 | 38.8% | 38.8% | Not listed | Not listed | 40.0% | Not listed | 42.5% | Not listed | Not listed | 48.1% | 28.1% | Not listed | 50.0% |
| SWE-bench Multilingual | 81.2% | 88.9% | Not listed | 79.9% | 82.9% | 77.5% | 87.6% | Not listed | 79.7% | 91.4% | 78.2% | 85.8% | 83.7% |
| SWE-bench Multimodal v2 | 43.1% | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | 41.3% | Not listed | 44.2% | 37.5% | 40.8% | 44.0% |
| SWE-bench Pro v2hard set | 86.3% | Not listed | Not listed | 84.3% | Not listed | 82.4% | 84.3% | Not listed | Not listed | 88.2% | Not listed | 58.8% | 52.9% |
| SWE-bench Pro v2full set | 97.7% | Not listed | Not listed | 96.6% | Not listed | 95.0% | 95.6% | Not listed | Not listed | 97.7% | Not listed | 94.9% | 87.2% |
| SWE-bench Verified | 82.2% | 92.0% | Not listed | 81.1% | 84.6% | 80.0% | 95.4% | Not listed | 80.0% | 93.4% | 76.1% | 80.0% | 88.9% |
| PostTrainBench v1.2 | 31.9% | 36.8% | Not listed | Not listed | 30.5% | Not listed | 34.5% | Not listed | Not listed | 31.5% | Not listed | Not listed | Not listed |
| BioMysteryBench | 70.0% | Not listed | 69.3% | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | 72.6% | 61.5% | 62.2% | 70.4% |
| BrowseCompwith context management | 90.2% | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | 84.5% | Not listed | Not listed | Not listed | Not listed |
| DeepSearchQA | 90.9% | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | 95.0% | Not listed | Not listed | Not listed |
| MCP Atlasfull set | 85.4% | Not listed | Not listed | Not listed | Not listed | 89.2% | 84.2% | Not listed | Not listed | 82.3% | Not listed | Not listed | Not listed |
| MCP Atlaspublic set | 85.0% | 82.0% | 62.9% | 80.4% | 83.2% | 87.5% | 83.6% | 84.4% | 84.5% | 84.2% | 77.8% | 87.5% | 81.0% |
| EnterpriseBench-CoreCraft | 71.8% | 46.7% | Not listed | Not listed | Not listed | Not listed | 51.8% | Not listed | Not listed | 43.1% | Not listed | 58.5% | Not listed |
| Finance Agent v2.1 | 67.5% | 60.6% | 63.3% | 58.8% | 62.9% | 60.7% | 65.0% | 64.3% | 61.7% | 60.0% | 56.8% | 65.8% | 65.4% |
| Excel Modeling Benchmark v1.1 | 69.5% | 66.3% | 68.9% | 67.8% | 69.2% | 62.0% | 69.1% | 67.4% | 68.7% | 71.3% | 70.0% | 72.9% | 65.5% |
| Tax Agent Bench | 65.8% | 62.0% | 59.9% | 62.7% | 74.9% | 52.1% | 73.1% | 64.9% | 67.5% | 68.7% | 58.9% | 66.8% | 62.2% |
| APEX-Accounting v1.1pass@1 | 14.5% | 12.3% | Not listed | 13.5% | 12.5% | 13.8% | 13.9% | Not listed | 12.0% | 9.5% | 10.8% | 15.5% | 15.3% |
| APEX-Accounting v1.1mean score | 62.0% | 53.7% | Not listed | 60.0% | 59.3% | 58.0% | 59.5% | Not listed | 56.5% | 56.7% | 53.9% | 58.8% | 62.2% |
| Diligence Benchbest harness | 58.1% | Not listed | Not listed | Not listed | 50.4% | 59.0% | 49.4% | Not listed | Not listed | Not listed | 42.1% | 51.1% | 40.8% |
| Harvey LAB v1.1 | 21.7% | 10.4% | 13.3% | 7.5% | 14.2% | 25.0% | 14.6% | 13.8% | 15.8% | 12.9% | 4.2% | 12.1% | 4.2% |
| Legal Research Bench | 46.6% | 45.2% | 38.0% | 34.1% | 45.2% | 38.5% | 49.0% | 47.1% | 35.6% | 46.2% | 30.3% | 38.9% | 43.8% |
| PerceptionBench | 64.7% | Not listed | Not listed | Not listed | 47.7% | Not listed | Not listed | Not listed | Not listed | 58.5% | 56.2% | Not listed | Not listed |
| ScreenSpot-Pro | 83.8% | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | 56.6% | 42.3% | 73.8% | 81.1% | Not listed | Not listed |
| ScreenSpot-Prowith Python | 91.5% | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed | 83.2% | 87.6% | Not listed | Not listed |
| BabyVision | 68.6% | 53.4% | Not listed | Not listed | Not listed | 44.3% | Not listed | Not listed | 42.3% | 53.4% | Not listed | Not listed | Not listed |
| BabyVisionwith Python | 93.7% | 82.6% | Not listed | 91.0% | 92.5% | 80.2% | Not listed | Not listed | 71.5% | 85.7% | Not listed | Not listed | Not listed |
| CharXiv (RQ) | 85.6% | 80.2% | 66.2% | 82.5% | 77.0% | 85.4% | Not listed | 74.2% | 83.2% | 84.8% | 81.8% | 86.2% | 84.4% |
| CharXiv (RQ)with Python | 87.7% | 89.4% | Not listed | Not listed | 85.6% | 88.7% | Not listed | Not listed | 87.9% | 91.3% | 85.3% | 92.1% | Not listed |
| Chartography | 36.7% | 16.3% | 7.5% | 12.6% | 17.9% | 35.7% | Not listed | 10.8% | 21.2% | 26.6% | 32.0% | 40.9% | 46.4% |
| Chartographywith Python | 80.4% | 78.0% | Not listed | Not listed | 83.0% | 63.2% | Not listed | Not listed | 56.6% | 68.1% | Not listed | 82.0% | Not listed |
| MathVision | 92.8% | 82.9% | Not listed | Not listed | Not listed | 87.7% | Not listed | Not listed | 88.9% | 93.1% | Not listed | Not listed | Not listed |
| SAGE v1.1 | 88.6% | 89.3% | 83.3% | 85.3% | Not listed | 85.8% | Not listed | 84.4% | Not listed | 86.8% | 83.4% | 88.2% | 89.8% |
| GPQA-Diamond | 93.0% | 91.2% | Not listed | 86.7% | 92.2% | 89.0% | 91.7% | 89.0% | 89.7% | 93.5% | Not listed | 95.3% | 89.8% |
| Apex 2025 | 63.5% | Not listed | Not listed | Not listed | Not listed | 32.3% | Not listed | Not listed | Not listed | 65.6% | Not listed | Not listed | Not listed |
| HMMT Feb 2026 | 95.8% | Not listed | Not listed | Not listed | Not listed | 93.9% | Not listed | Not listed | Not listed | 97.0% | Not listed | Not listed | Not listed |
| LongBench v2 | 70.6% | 68.2% | 60.7% | 60.9% | 69.0% | 68.7% | 68.4% | 68.0% | 65.7% | 69.0% | 65.0% | 73.7% | 65.3% |
| ComplexConstraints v1.0 | 44.5% | 27.6% | 22.1% | 23.2% | 50.4% | 41.3% | 30.5% | 32.3% | 36.5% | 37.9% | 27.3% | 48.4% | 36.8% |
| CyberBench v1.1 | 72.7% | 67.9% | 75.4% | Not listed | Not listed | Not listed | 72.1% | 72.9% | Not listed | 75.2% | 76.3% | 43.8% | 75.5% |
| PatchEval | 79.6% | 74.8% | Not listed | Not listed | 80.4% | 77.8% | 69.6% | Not listed | 76.1% | Not listed | Not listed | Not listed | Not listed |
Thinking Machines' table from the Inkling 2 Small model card, run at effort 0.99 (xhigh) with a 1M-token trajectory limit; where a benchmark's owner publishes scores (Datacurve, Scale AI, Surge, Mercor, Thoughtful, Vals), Thinking Machines uses them. DeepSeek-V4.1-Flash is the 1017 build. Left out here: the Artificial Analysis and Vals indexes, the rows Artificial Analysis ran (FrontierCode, Terminal-Bench, GDP.pdf, MMMU-Pro, HLE, CritPt, AA-LCR, AA-Omniscience, AutomationBench, EnterpriseOps-Gym, Tau 3 Banking), Arena and DesignArena ratings, PMPP-Hard (not a percentage), the safety rows, and JobBench, DAYJOB and Humanity's Sixth Sense, which we could not confirm are public.
Fit
Strengths, cautions and best uses
- Strengths
- Thinking Machines Lab's second model generation: a mixture-of-experts model with 594B total and 17B active parameters that reads images and takes a variable effort setting
- In its lab's table no other listed model scores higher on EnterpriseBench-CoreCraft (71.8%), Finance Agent v2.1 (67.5%) or ScreenSpot-Pro (83.8%)
- $0.30 input, $1.00 output and $0.03 cached input per 1M tokens on Tinker, which can also fine-tune it with RL or supervised training
- Cautions
- Closed weights for now: Thinking Machines says it will publish the full weights under Apache 2.0 the week after launch, once external safety testing ends; until then Tinker is the only way to use it
- Tinker serves it with a 64K or 256K context, although the model card lists up to 1M tokens
- In its lab's own table DeepSeek-V4.1-Flash scores higher on DeepSWE v1.1 (77.0% against 72.1%) and Claude Haiku 5.5 on Vibe Code Bench (90.4% against 81.6%); Thinking Machines also says it is less tuned for chat, and no independent benchmark score is recorded here yet
- Best for
- Agentic coding and office, finance and legal work at a low per-token price
- Teams that want to post-train their own variant on Tinker
From benchmark to production
Building on Inkling 2 Small?
Public scores rarely hold on your own data. We test it against the alternatives on your real tasks, then build the agent or pipeline around the model that wins.
AI Model Cost AuditLaunch price $100, regular price $150What the audit includes
Alternatives
Alternatives to Inkling 2 Small
The closest current text models by price. The cheapest model that scores as well or better is listed first; failing that, the cheapest one within 3% of the benchmark's scale.
On score and price
Inkling 2 Small beside the 4 alternatives listed below, on Arena Text and on blended price per 1M tokens.
Arena Text
Measured October 28, 2026, higher is better
- Inkling 2 Small
- Not scored
- DeepSeek-V4.1-Flash
- 1,476
- Gemini 3.5 Flash-Lite
- 1,456
- LongCat-2.5
- Not scored
- Solar Pro 4
- 1,388
Blended price per 1M tokens
Listed today, lower is better
- Inkling 2 Small
- $0.475
- DeepSeek-V4.1-Flash
- $0.525
- Gemini 3.5 Flash-Lite
- $0.85
- LongCat-2.5
- $0.525
- Solar Pro 4
- $0.525
DeepSeek-V4.1-Flash
- Same price per 1M input tokens, $0.30
- 1M context window vs 262K
- From DeepSeek
Gemini 3.5 Flash-Lite
- Same price per 1M input tokens, $0.30
- 1.05M context window vs 262K
- From Google
LongCat-2.5
- Same price per 1M input tokens, $0.30
- 1.05M context window vs 262K
- From Meituan
Solar Pro 4
- Same price per 1M input tokens, $0.30
- From Upstage
Further reading
More on Inkling 2 Small
Compare Inkling 2 Small with
- Inkling 2 Small vs Claude Haiku 5.5
- Inkling 2 Small vs DeepSeek-V4.1-Flash
- Inkling 2 Small vs Gemini 3.5 Flash-Lite
- Inkling 2 Small vs Gemini 3.8 Flash
- Inkling 2 Small vs GLM-5.3-Flash
- Inkling 2 Small vs GLM-5.3
- Inkling 2 Small vs GPT-6 Luna
- Inkling 2 Small vs Kimi K3
- Inkling 2 Small vs LongCat-2.5
- Inkling 2 Small vs MiMo-V2.6-Flash
- Inkling 2 Small vs MiniMax M3.1 Flash
- Inkling 2 Small vs Muse Spark Open
- Inkling 2 Small vs Solar Pro 4
FAQ
Inkling 2 Small FAQ
How much does Inkling 2 Small cost?
Inkling 2 Small costs $0.30 per 1M input tokens and $1 per 1M output tokens at the standard tier. It is the #11 cheapest of 58 current text models priced per 1M input tokens. Source: Thinking Machines Lab, verified November 6, 2026.
What is the context window of Inkling 2 Small?
Inkling 2 Small accepts up to 262,144 tokens of context. That ranks #57 of 67 current text models by context length.
What benchmark results has Thinking Machines Lab published for Inkling 2 Small?
In the Inkling 2 Small model card (November 5, 2026), Thinking Machines Lab reports 72.1% on DeepSWE v1.1, 81.6% on Vibe Code Bench v1.1, 14.1% on Sudo L7 and 38.8% on SWE-Marathon v1.1. These are Thinking Machines Lab's own figures, beside the rivals it chose, not independent measurements.
Is Inkling 2 Small open source or API only?
Inkling 2 Small is available through Thinking Machines Lab's API only; its weights are not published.
What are the best alternatives to Inkling 2 Small?
The closest current alternatives by price and measured performance are DeepSeek-V4.1-Flash (same price per 1M input tokens, $0.30), Gemini 3.5 Flash-Lite (same price per 1M input tokens, $0.30), LongCat-2.5 (same price per 1M input tokens, $0.30) and Solar Pro 4 (same price per 1M input tokens, $0.30).
Sources
Sources
- Introducing Inkling 2 Small · Thinking Machines Lab · official · verified November 6, 2026
- Inkling 2 Small model card · Thinking Machines Lab · official · verified November 6, 2026
- Tinker models and pricing · Thinking Machines Lab · official · verified November 6, 2026
All 255 sources behind the benchmarks, with their versions, licenses and notes, are in the source catalog.
Cite this page
Van Data Team. "Inkling 2 Small Benchmarks, Pricing and Specs." AI Model Benchmark Dataset 2026-11-06. https://vandatateam.com/ai-model-benchmarks/inkling-2-small
Reuse the data: JSON · CSV · or explore every model in the AI model benchmark comparison.
Production evaluation
Deciding whether Inkling 2 Small fits your workload?
Public benchmarks narrow the shortlist. We test the finalists on your own tasks, data and cost limits, then build the routing and review controls to run the winner in production.
- Evaluation cases and acceptance thresholds tied to real work
- Model routing, fallbacks and cost controls