Thinking Machines Lab model product
Inkling
Inkling is a general-purpose multimodal model that accepts text, image, and audio inputs and generates text outputs.
Updated Sep 25, 2026. Default version: Inkling
LLMBoard Score63.6Inkling
Coverage100%16 benchmark families
Context window524.3KTokens
Official input priceN/AOfficial price unavailable
Inkling Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Inkling Benchmark Results
Benchmark scores for Inkling.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkMMAU | Score77.20% | Rank01 | Participants4 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkSimpleQA Verified | Score43.90% | Rank01 | Participants4 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkVoiceBench Avg | Score91.40% | Rank01 | Participants3 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkAA-Omniscience Index | Score2.10 points | Rank02 | Participants3 | Percentile50.00% | EvidenceC | Evaluated |
| BenchmarkAIME 2026 | Score97.10% | Rank02 | Participants26 | Percentile96.00% | EvidenceC | Evaluated |
| BenchmarkGlobal-MMLU-Lite | Score88.70% | Rank02 | Participants16 | Percentile93.33% | EvidenceC | Evaluated |
| BenchmarkTau3 Banking | Score23.70% | Rank05 | Participants11 | Percentile60.00% | EvidenceC | Evaluated |
| BenchmarkHumanity's Last Exam (no tools, text-only) | Score29.70% | Rank07 | Participants8 | Percentile14.29% | EvidenceC | Evaluated |
| BenchmarkIFBench | Score79.80% | Rank08 | Participants42 | Percentile82.93% | EvidenceC | Evaluated |
| BenchmarkHumanity's Last Exam (with tools, text-only) | Score46.00% | Rank09 | Participants10 | Percentile11.11% | EvidenceC | Evaluated |
| BenchmarkGDPval-AA | Score1,238.00 points | Rank11 | Participants12 | Percentile9.09% | EvidenceC | Evaluated |
| BenchmarkMCP Atlas | Score76.00% | Rank14 | Participants36 | Percentile62.86% | EvidenceC | Evaluated |
| BenchmarkLiveBench math (2026-06-25) | Score88.36 score | Rank21 | Participants41 | Percentile50.00% | EvidenceB | EvaluatedN/A |
| BenchmarkLM Arena Agent Bash Recovery Steps | Score1.96% | Rank22 | Participants42 | Percentile48.78% | EvidenceA | Evaluated |
| BenchmarkLiveBench instruction (2026-06-25) | Score70.10 score | Rank22 | Participants41 | Percentile47.50% | EvidenceB | EvaluatedN/A |
| BenchmarkSWE-Bench Verified | Score77.60% | Rank27 | Participants116 | Percentile77.39% | EvidenceC | Evaluated |
| BenchmarkLM Arena Agent Tool Hallucination | Score-0.03% | Rank30 | Participants42 | Percentile29.27% | EvidenceA | Evaluated |
| BenchmarkCharXiv-R | Score78.10% | Rank31 | Participants58 | Percentile47.37% | EvidenceC | Evaluated |
| BenchmarkAA Omniscience Accuracy | Score41.55% | Rank31 | Participants201 | Percentile85.00% | EvidenceB | EvaluatedN/A |
| BenchmarkTerminal-Bench 2.1 | Score63.80% | Rank32 | Participants42 | Percentile24.39% | EvidenceC | Evaluated |
| BenchmarkLiveBench reasoning (2026-06-25) | Score78.35 score | Rank32 | Participants41 | Percentile22.50% | EvidenceB | EvaluatedN/A |
| BenchmarkMMMU-Pro | Score73.50% | Rank38 | Participants72 | Percentile47.89% | EvidenceC | Evaluated |
| BenchmarkLM Arena Agent Leaderboard | Score-10.02% | Rank38 | Participants42 | Percentile9.76% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Steerability | Score-10.51% | Rank40 | Participants42 | Percentile4.88% | EvidenceA | Evaluated |
| BenchmarkLM Arena Agent Task Outcome Explicit | Score-19.32% | Rank40 | Participants42 | Percentile4.88% | EvidenceA | Evaluated |
| BenchmarkAA SciCode Subtasks | Score46.99% | Rank40 | Participants89 | Percentile55.68% | EvidenceB | EvaluatedN/A |
| BenchmarkLM Arena Agent Praise Complaint | Score-22.20% | Rank42 | Participants42 | Percentile0.00% | EvidenceA | Evaluated |
| BenchmarkAA HLE Text No Tools | Score31.88% | Rank46 | Participants200 | Percentile77.39% | EvidenceB | EvaluatedN/A |
| BenchmarkLM Arena Text Factuality | Score1,449.67 rating | Rank47 | Participants125 | Percentile62.90% | EvidenceA | Evaluated |
| BenchmarkAA CritPt | Score5.43% | Rank47 | Participants200 | Percentile76.88% | EvidenceB | EvaluatedN/A |
Inkling Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenatext factuality | Categorymath | Rank09 | Rating / score1,487.33 | Votes881 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryspanish | Rank18 | Rating / score1,459.48 | Votes478 | ObservationsN/A | Result date |
| Arenatext | Categorymath | Rank19 | Rating / score1,477.41 | Votes1,120 | ObservationsN/A | Result date |
| Arenatext | Categorypolish | Rank21 | Rating / score1,464.38 | Votes403 | ObservationsN/A | Result date |
| Arenaagent bash recovery steps | Categoryoverall | Rank22 | Rating / score0.02 | VotesN/A | Observations45.1K | Result date |
| Arenatext factuality | Categoryindustry mathematical | Rank23 | Rating / score1,466.97 | Votes1,088 | ObservationsN/A | Result date |
| Arenatext style control | Categorymath | Rank24 | Rating / score1,476.88 | Votes1,120 | ObservationsN/A | Result date |
| Arenaagent tool hallucination | Categoryoverall | Rank30 | Rating / score0.00 | VotesN/A | Observations1.5M | Result date |
| Arenatext factuality | Categoryindustry legal and government | Rank34 | Rating / score1,469.16 | Votes1,748 | ObservationsN/A | Result date |
| Arenatext style control | Categorypolish | Rank35 | Rating / score1,460.40 | Votes403 | ObservationsN/A | Result date |
| Arenatext factuality | Categorychinese | Rank37 | Rating / score1,495.02 | Votes1,483 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryexpert | Rank37 | Rating / score1,484.23 | Votes2,595 | ObservationsN/A | Result date |
| Arenaagent | Categoryoverall | Rank38 | Rating / score-0.10 | VotesN/A | Observations1.6M | Result date |
| Arenatext factuality | Categoryindustry business and management and financial operations | Rank39 | Rating / score1,457.34 | Votes4,718 | ObservationsN/A | Result date |
| Arenaagent steerability | Categoryoverall | Rank40 | Rating / score-0.11 | VotesN/A | Observations30.9K | Result date |
| Arenaagent task outcome explicit | Categoryoverall | Rank40 | Rating / score-0.19 | VotesN/A | Observations24.1K | Result date |
| Arenatext | Categoryjapanese | Rank40 | Rating / score1,412.38 | Votes408 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryindustry software and it services | Rank40 | Rating / score1,490.46 | Votes10,436 | ObservationsN/A | Result date |
| Arenatext | Categorykorean | Rank41 | Rating / score1,401.90 | Votes500 | ObservationsN/A | Result date |
| Arenaagent praise complaint | Categoryoverall | Rank42 | Rating / score-0.22 | VotesN/A | Observations9.4K | Result date |
| Arenatext | Categoryindustry mathematical | Rank42 | Rating / score1,459.85 | Votes1,369 | ObservationsN/A | Result date |
| Arenatext factuality | Categorynon english | Rank42 | Rating / score1,439.32 | Votes15,006 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry business and management and financial operations | Rank43 | Rating / score1,438.57 | Votes5,051 | ObservationsN/A | Result date |
| Arenatext | Categoryexpert | Rank44 | Rating / score1,464.45 | Votes2,971 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryindustry life and physical and social science | Rank45 | Rating / score1,472.45 | Votes3,847 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryindustry medicine and healthcare | Rank45 | Rating / score1,472.79 | Votes1,450 | ObservationsN/A | Result date |
| Arenatext | Categorychinese | Rank46 | Rating / score1,490.07 | Votes1,820 | ObservationsN/A | Result date |
| Arenatext | Categorynon english | Rank46 | Rating / score1,430.27 | Votes15,116 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry software and it services | Rank47 | Rating / score1,464.96 | Votes10,523 | ObservationsN/A | Result date |
| Arenatext factuality | Categorycoding | Rank47 | Rating / score1,499.77 | Votes7,297 | ObservationsN/A | Result date |
Inkling Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $0.95 input, $4.05 output per 1M via Deep Infra
- Tracked offerings
- 21
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderNvidia | Provider model IDthinkingmachines/inkling | Regionglobal | Input / 1MN/A | Output / 1MN/A | Context1M | Updated |
| ProviderDeep Infra | Provider model IDthinkingmachines/Inkling | Regionglobal | Input / 1M$0.95 | Output / 1M$4.05 | Context524.3K | Updated |
| ProviderKilo Gateway | Provider model IDthinkingmachines/inkling | Regionglobal | Input / 1M$0.95 | Output / 1M$4.05 | Context524.3K | Updated |
| ProviderDevPass (LLM Gateway) | Provider model IDinkling | Regionglobal | Input / 1M$0.95 | Output / 1M$4.05 | Context524.3K | Updated |
| ProviderNanoGPT | Provider model IDthinkingmachines/inkling | Regionglobal | Input / 1M$1 | Output / 1M$4.05 | Context1M | Updated |
| ProviderHugging Face | Provider model IDthinkingmachines/Inkling | Regionglobal | Input / 1M$1 | Output / 1M$4.05 | Context1M | Updated |
| ProviderOpenRouter | Provider model IDthinkingmachines/inkling | Regionglobal | Input / 1M$1 | Output / 1M$4.05 | Context1M | Updated |
| ProviderMerge Gateway | Provider model IDthinkingmachines/inkling | Regionglobal | Input / 1M$1 | Output / 1M$4.05 | Context1M | Updated |
| ProviderTogether AI | Provider model IDthinkingmachines/Inkling | Regionglobal | Input / 1M$1 | Output / 1M$4.05 | Context524.3K | Updated |
| ProviderNeon | Provider model IDinkling | Regionglobal | Input / 1M$1 | Output / 1M$4.05 | Context1M | Updated |
| ProviderBaseten | Provider model IDthinkingmachines/inkling | Regionglobal | Input / 1M$1 | Output / 1M$4.05 | Context1M | Updated |
| ProviderVercel AI Gateway | Provider model IDthinkingmachines/inkling | Regionglobal | Input / 1M$1 | Output / 1M$4.05 | Context256K | Updated |
| ProviderFireworks AI | Provider model IDaccounts/fireworks/models/inkling | Regionglobal | Input / 1M$1 | Output / 1M$4.05 | Context1M | Updated |
| ProviderCharm Hyper | Provider model IDinkling | Regionglobal | Input / 1M$1.09 | Output / 1M$4.41 | Context1M | Updated |
| ProviderModal | Provider model IDthinkingmachines/Inkling-NVFP4 | Regionglobal | Input / 1M$1.2 | Output / 1M$5 | Context1M | Updated |
| ProviderVenice AI | Provider model IDinkling | Regionglobal | Input / 1M$1.25 | Output / 1M$5.06 | Context524.3K | Updated |
| ProviderImpossibl | Provider model IDthinkingmachines/inkling | Regionglobal | Input / 1M$1.87 | Output / 1M$4.68 | Context65.5K | Updated |
| ProviderLLMTR | Provider model IDthinkingmachines/inkling | Regionglobal | Input / 1M$1.87 | Output / 1M$4.68 | Context262.1K | Updated |
| ProviderRequesty | Provider model IDinkling | Regionglobal | Input / 1M$1.87 | Output / 1M$4.68 | Context65.5K | Updated |
| ProviderThinking Machines | Provider model IDthinkingmachines/Inkling | Regionglobal | Input / 1M$1.87 | Output / 1M$4.68 | Context65.5K | Updated |
| ProviderAbacus | Provider model IDthinkingmachines/Inkling | Regionglobal | Input / 1M$3.74 | Output / 1M$9.36 | Context262.1K | Updated |
Inkling Runtime Performance
Provider-specific output speed and catalog latency for Inkling. 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 Specifications
Technical details for the model's default version.
- Version
- Inkling
- Released
- Jul 21, 2026
- Knowledge cutoff
- Unknown
- Parameters
- 975B
- Context window
- 524.3K
- Max output
- 524.3K
- Inputs
- audio, image, text
- Outputs
- text
- Open weights
- Yes
- License
- Apache 2.0
Inkling 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 | Released | LLMBoard63.57 | Parameters975B | Context524.3K | Max output524.3K | Open weightsYes | LicenseApache 2.0 |
Models similar to Inkling
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
TM
Inkling Small
Thinking Machines Lab
GO
Google
Gemini 3.8 Flash Cyber
AC
Qwen3.5 397B A17B
Alibaba Cloud / Qwen Team
GOGoogle
Gemini 3 Flash
MA
Kimi K2.5
Moonshot AI
MA
Kimi K2.7 Code
Moonshot AI
What is Inkling?
Key information about Inkling and its available data.
Inkling is a general-purpose multimodal model from Thinking Machines Lab. It accepts text, image, and audio inputs and generates text outputs.
Data as of 2026-09-24.
FAQ
Common questions about Inkling.
When was Inkling released?
Inkling's default version was released on Jul 21, 2026.
How much does Inkling cost?
No official standard PAYG price is currently available for Inkling. The lowest tracked third-party offer starts at $0.95 input and $4.05 output via Deep Infra.
Who created Inkling?
Inkling was created by Thinking Machines Lab.
What is the context window for Inkling?
The default version has a 524.3K token context window.
Is Inkling open weight?
Yes. The default version is marked as open weight under Apache 2.0.
How many API providers offer Inkling?
21 provider offerings are linked to the default version.