Thinking Machines Lab model product
Inkling Small
Inkling-Small is an open-weights multimodal MoE model from Thinking Machines Lab with 276B total and 12B active parameters.
Updated Oct 23, 2026. Default version: Inkling-Small
LLMBoard Score62.9Inkling-Small
Coverage80%24 benchmark families
Context window524.3KTokens
Official input priceN/AOfficial price unavailable
Inkling Small Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Inkling Small Benchmark Results
Benchmark scores for Inkling-Small.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkIFBench | Score82.20% | Rank02 | Participants46 | Percentile97.78% | EvidenceC | Evaluated |
| BenchmarkMMAU | Score77.00% | Rank03 | Participants5 | Percentile50.00% | EvidenceC | Evaluated |
| BenchmarkVoiceBench Avg | Score90.10% | Rank03 | Participants4 | Percentile33.33% | EvidenceC | Evaluated |
| BenchmarkAA-Briefcase | Score917.00 points | Rank04 | Participants4 | Percentile0.00% | EvidenceC | Evaluated |
| BenchmarkSimpleQA Verified | Score20.60% | Rank04 | Participants5 | Percentile25.00% | EvidenceC | Evaluated |
| BenchmarkCritPT | Score8.30% | Rank05 | Participants7 | Percentile33.33% | EvidenceC | Evaluated |
| BenchmarkGlobal-MMLU-Lite | Score86.70% | Rank05 | Participants16 | Percentile73.33% | EvidenceC | Evaluated |
| BenchmarkAIME 2026 | Score95.50% | Rank06 | Participants27 | Percentile80.77% | EvidenceC | Evaluated |
| BenchmarkARC-AGI | Score84.00% | Rank07 | Participants12 | Percentile45.45% | EvidenceC | Evaluated |
| BenchmarkMCP Atlas | Score79.60% | Rank10 | Participants39 | Percentile76.32% | EvidenceC | Evaluated |
| BenchmarkSciCode | Score48.70% | Rank10 | Participants31 | Percentile70.00% | EvidenceC | Evaluated |
| BenchmarkARC-AGI v2 | Score40.10% | Rank11 | Participants19 | Percentile44.44% | EvidenceC | Evaluated |
| BenchmarkGDPval-AA | Score1,269.00 points | Rank11 | Participants17 | Percentile37.50% | EvidenceC | Evaluated |
| BenchmarkArtificial Analysis | Score40.00% | Rank12 | Participants13 | Percentile8.33% | EvidenceC | Evaluated |
| BenchmarkTau3 Banking | Score15.50% | Rank12 | Participants16 | Percentile26.67% | EvidenceC | Evaluated |
| BenchmarkSWE-Bench Verified | Score80.20% | Rank14 | Participants120 | Percentile89.08% | EvidenceC | Evaluated |
| BenchmarkLM Arena Agent Bash Recovery Steps | Score5.43% | Rank14 | Participants44 | Percentile69.77% | EvidenceA | Evaluated |
| BenchmarkToolathlon | Score54.40% | Rank16 | Participants42 | Percentile63.41% | EvidenceC | Evaluated |
| BenchmarkLM Arena Agent Bash Reliability No Prereq | Score4.97% | Rank17 | Participants45 | Percentile63.64% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Bash Reliability | Score2.90% | Rank18 | Participants44 | Percentile60.47% | EvidenceA | Evaluated |
| BenchmarkBrowseComp | Score77.40% | Rank26 | Participants70 | Percentile63.77% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score89.50% | Rank33 | Participants256 | Percentile87.45% | EvidenceC | Evaluated |
| BenchmarkAA SciCode Subtasks | Score49.65% | Rank33 | Participants89 | Percentile63.64% | EvidenceB | EvaluatedN/A |
| BenchmarkTerminal-Bench 2.1 | Score64.70% | Rank35 | Participants48 | Percentile27.66% | EvidenceC | Evaluated |
| BenchmarkMMMU-Pro | Score74.00% | Rank36 | Participants73 | Percentile51.39% | EvidenceC | Evaluated |
| BenchmarkLM Arena Agent Tool Hallucination | Score-1.49% | Rank37 | Participants44 | Percentile16.28% | EvidenceA | Evaluated |
| BenchmarkCharXiv-R | Score77.40% | Rank38 | Participants60 | Percentile37.29% | EvidenceC | Evaluated |
| BenchmarkAA GPQA Diamond | Score89.49% | Rank41 | Participants199 | Percentile79.80% | EvidenceB | EvaluatedN/A |
| BenchmarkAA CritPt | Score8.29% | Rank41 | Participants200 | Percentile79.90% | EvidenceB | EvaluatedN/A |
| BenchmarkSWE-Bench Pro | Score55.90% | Rank42 | Participants63 | Percentile33.87% | EvidenceC | Evaluated |
Inkling Small Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenaagent bash recovery steps | Categoryoverall | Rank14 | Rating / score0.05 | VotesN/A | Observations15.4K | Result date |
| Arenaagent bash reliability no prereq | Categoryoverall | Rank17 | Rating / score0.05 | VotesN/A | Observations7.4K | Result date |
| Arenaagent bash reliability | Categoryoverall | Rank18 | Rating / score0.03 | VotesN/A | Observations12.2K | Result date |
| Arenatext factuality | Categorymath | Rank36 | Rating / score1,455.96 | Votes1,226 | ObservationsN/A | Result date |
| Arenaagent tool hallucination | Categoryoverall | Rank37 | Rating / score-0.01 | VotesN/A | Observations473.9K | Result date |
| Arenaagent | Categoryoverall | Rank43 | Rating / score-0.16 | VotesN/A | Observations520.3K | Result date |
| Arenaagent task outcome explicit | Categoryoverall | Rank43 | Rating / score-0.27 | VotesN/A | Observations7.3K | Result date |
| Arenawebdev | Categoryimage to webdev | Rank43 | Rating / score1,410.02 | Votes1,025 | ObservationsN/A | Result date |
| Arenaagent praise complaint | Categoryoverall | Rank44 | Rating / score-0.22 | VotesN/A | Observations3.2K | Result date |
| Arenaagent steerability | Categoryoverall | Rank44 | Rating / score-0.22 | VotesN/A | Observations23.7K | Result date |
| Arenavision | Categorychinese | Rank45 | Rating / score1,292.40 | Votes461 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryindustry mathematical | Rank46 | Rating / score1,449.55 | Votes1,516 | ObservationsN/A | Result date |
| Arenatext | Categorymath | Rank48 | Rating / score1,458.89 | Votes1,556 | ObservationsN/A | Result date |
| Arenavision | Categoryhomework | Rank56 | Rating / score1,257.44 | Votes835 | ObservationsN/A | Result date |
| Arenatext style control | Categorymath | Rank58 | Rating / score1,451.58 | Votes1,556 | ObservationsN/A | Result date |
| Arenavision style control | Categorychinese | Rank58 | Rating / score1,245.09 | Votes461 | ObservationsN/A | Result date |
| Arenavision | Categorydiagram | Rank59 | Rating / score1,250.72 | Votes2,071 | ObservationsN/A | Result date |
| Arenavision | Categoryenglish | Rank59 | Rating / score1,242.61 | Votes2,816 | ObservationsN/A | Result date |
| Arenavision | Categoryhumor | Rank59 | Rating / score1,207.39 | Votes384 | ObservationsN/A | Result date |
| Arenavision | Categoryoverall | Rank60 | Rating / score1,236.80 | Votes8,074 | ObservationsN/A | Result date |
| Arenatext | Categoryjapanese | Rank61 | Rating / score1,401.01 | Votes569 | ObservationsN/A | Result date |
| Arenavision | Categoryocr | Rank61 | Rating / score1,242.91 | Votes5,655 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryfrench | Rank63 | Rating / score1,445.43 | Votes830 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryspanish | Rank63 | Rating / score1,422.28 | Votes755 | ObservationsN/A | Result date |
| Arenawebdev | Categorywebdev-react | Rank63 | Rating / score1,399.28 | Votes3,554 | ObservationsN/A | Result date |
| Arenavision style control | Categoryhomework | Rank65 | Rating / score1,244.43 | Votes835 | ObservationsN/A | Result date |
| Arenavision | Categorycreative writing vision | Rank66 | Rating / score1,205.16 | Votes675 | ObservationsN/A | Result date |
| Arenavision style control | Categorydiagram | Rank66 | Rating / score1,238.07 | Votes2,071 | ObservationsN/A | Result date |
| Arenavision style control | Categoryhumor | Rank66 | Rating / score1,171.48 | Votes384 | ObservationsN/A | Result date |
| Arenawebdev | Categoryoverall | Rank66 | Rating / score1,409.37 | Votes5,323 | ObservationsN/A | Result date |
Inkling Small Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $0.30 input, $1.2 output per 1M via Thinking Machines
- Tracked offerings
- 15
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderThinking Machines | Provider model IDthinkingmachines/Inkling-Small:peft:262144:sampling-nvfp4 | Regionglobal | Input / 1M$0.30 | Output / 1M$1.2 | Context262.1K | Updated |
| ProviderNanoGPT | Provider model IDthinkingmachines/Inkling-Small | Regionglobal | Input / 1M$0.45 | Output / 1M$1.2 | Context524.3K | Updated |
| ProviderDeep Infra | Provider model IDthinkingmachines/Inkling-Small | Regionglobal | Input / 1M$0.45 | Output / 1M$1.2 | Context524.3K | Updated |
| ProviderOpenRouter | Provider model IDthinkingmachines/inkling-small | Regionglobal | Input / 1M$0.45 | Output / 1M$1.2 | Context524.3K | Updated |
| ProviderVercel AI Gateway | Provider model IDthinkingmachines/inkling-small | Regionglobal | Input / 1M$0.45 | Output / 1M$1.2 | Context1M | Updated |
| ProviderKilo Gateway | Provider model IDthinkingmachines/inkling-small | Regionglobal | Input / 1M$0.45 | Output / 1M$1.2 | Context524.3K | Updated |
| ProviderHugging Face | Provider model IDthinkingmachines/Inkling-Small | Regionglobal | Input / 1M$0.50 | Output / 1M$1.2 | Context524.3K | Updated |
| ProviderArcee | Provider model IDthinkingmachines/inkling-small | Regionglobal | Input / 1M$0.50 | Output / 1M$1.2 | Context262.1K | Updated |
| ProviderBaseten | Provider model IDthinkingmachines/inkling-small | Regionglobal | Input / 1M$0.50 | Output / 1M$1.2 | Context1M | Updated |
| ProviderPioneer | Provider model IDthinkingmachines/inkling-small | Regionglobal | Input / 1M$0.50 | Output / 1M$1.2 | Context1M | Updated |
| ProviderCommandCode | Provider model IDthinkingmachines/inkling-small | Regionglobal | Input / 1M$0.50 | Output / 1M$1.2 | Context1M | Updated |
| ProviderLLMTR | Provider model IDthinkingmachines/inkling-small | Regionglobal | Input / 1M$0.58 | Output / 1M$1.44 | Context262.1K | Updated |
| ProviderThinking Machines | Provider model IDthinkingmachines/Inkling-Small:peft:262144 | Regionglobal | Input / 1M$0.58 | Output / 1M$1.44 | Context262.1K | Updated |
| ProviderThinking Machines | Provider model IDthinkingmachines/Inkling-Small | Regionglobal | Input / 1M$0.58 | Output / 1M$1.44 | Context65.5K | Updated |
| ProviderAI/ML API | Provider model IDthinkingmachines/inkling-small | Regionglobal | Input / 1M$0.6136 | Output / 1M$1.52 | Context524.3K | Updated |
Inkling Small Runtime Performance
Provider-specific output speed and catalog latency for Inkling-Small. Runtime does not affect the capability score.
No runtime data
No provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Inkling Small Specifications
Technical details for the model's default version.
- Version
- Inkling-Small
- Released
- Jul 30, 2026
- Knowledge cutoff
- Unknown
- Parameters
- 276B
- Context window
- 524.3K
- Max output
- 524.3K
- Inputs
- audio, image, text
- Outputs
- text
- Open weights
- Yes
- License
- Apache 2.0
Inkling Small Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionInkling-Small | Released | LLMBoard62.91 | Parameters276B | Context524.3K | Max output524.3K | Open weightsYes | LicenseApache 2.0 |
Models similar to Inkling Small
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
TM
Inkling
Thinking Machines Lab
GO
Google
Gemini 3.8 Flash Cyber
AC
Qwen3.8 Omni Flash
Alibaba Cloud / Qwen Team
AC
Qwen3.5 397B A17B
Alibaba Cloud / Qwen Team
GOGoogle
Gemini 3 Flash
MA
Kimi K2.7 Code
Moonshot AI
What is Inkling Small?
Key information about Inkling Small and its available data.
Inkling-Small is a model from Thinking Machines Lab that accepts text, image, and audio inputs and generates text. 0 and supports a context window of up to 1M tokens, with native reasoning and variable thinking effort.
Data as of 2026-10-22.
FAQ
Common questions about Inkling Small.
When was Inkling Small released?
Inkling Small's default version was released on Jul 30, 2026.
How much does Inkling Small cost?
No official standard PAYG price is currently available for Inkling Small. The lowest tracked third-party offer starts at $0.30 input and $1.2 output via Thinking Machines.
Who created Inkling Small?
Inkling Small was created by Thinking Machines Lab.
What is the context window for Inkling Small?
The default version has a 524.3K token context window.
Is Inkling Small open weight?
Yes. The default version is marked as open weight under Apache 2.0.
How many API providers offer Inkling Small?
15 provider offerings are linked to the default version.