The free Inkling Small endpoint is only available for use with agentic harnesses. Do not upload any confidential information or personal data (e.g., voices and images of people's faces). Your usage of this free endpoint, including prompts and outputs, is logged and used to improve Thinking Machines Lab's models, products, and services. The logged session data will be disassociated from your account and other persistent identifiers before being used for these purposes.
By using this free endpoint, you agree to the TML Free Research API Terms of ServiceOpens in new tab. For more information about Thinking Machines Lab's data processing practices, see this Privacy NoticeOpens in new tab.
Thinking Machines: Inkling Small
thinkingmachines/inkling-small
Inkling Small is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. It is positioned as the smaller, more efficient member of the Inkling family and is suited for reasoning, coding, agentic workflows, retrieval-augmented generation, instruction following, and multilingual conversation.
Providers
This model is hosted by one provider. OpenRouter forwards every request without routing.
| $0.45 | $1.20 | $0.10 | 0.29s | 94 tok/s |
Input $0.45 / Output $1.2094 tok/s99.74% uptime
Pricing
The average price customers actually pay for this model, next to the prices providers post. Caching and discounts mean the price actually paid is often well below the listed one.
Weighted average
Inkling Small — Price history
| $0.136 | $1.20 | 89.6% | 100.0% |
- Effective in $0.136 / Effective out $1.20Token share100.0%
Performance
Throughput is how fast the model writes (tokens per second — higher is better). Latency is total round-trip time (lower is better). TTFT is time-to-first-token — how long before you see anything appear (lower is better).
P50, best across providers
P50, best provider
Uptime
Uptime is the percentage of the past 3 days that at least one provider was responding to requests. Availability is the percentage of time that inference was successfully served. OpenRouter continuously monitors and uses the next-best provider when one returns an error.
Availability over the last 3 days
Last 72 hoursAvailability over the last 24 hours
When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it. You can access per-provider uptime data programmatically through the Endpoints API. Learn more about our load balancing and provider routing options.
Benchmarks
Scores on standardized evaluations and this model’s own outputs on shared benchmark prompts. Higher percentages are better — and each rank shows where this model lands among the models evaluated on that source.
| Source | Benchmark | Score |
|---|---|---|
| Artificial Analysis | Inkling Small Intelligence Index | 12.9 |
| Artificial Analysis | Inkling Small Finance & Accounting Index | 23.7 |
| Artificial Analysis | Inkling Small Strategy & Ops Index | 18.0 |
| Artificial Analysis | Inkling Small Legal Index | 21.7 |
| Artificial Analysis | Inkling Small Healthcare & Medical Index | 18.6 |
| Artificial Analysis | Inkling Small Engineering Index | 23.1 |
| Artificial Analysis | Inkling Small Economics Index | 24.9 |
| Artificial Analysis | Inkling Small Openness Index | 33.3 |
| Artificial Analysis | Inkling Small AutomationBench-AA | 4.8% |
| Artificial Analysis | Inkling Small GDP.pdf | 11.2% |
| Artificial Analysis | Inkling Small AA-AnalystAgent | 27.5% |
| Artificial Analysis | Inkling Small EnterpriseOps-Gym-AA | 42.7% |
| Artificial Analysis | Inkling Small Harvey LAB-AA | 0.0% |
| Artificial Analysis | Inkling Small GPQA Diamond | 89.5% |
| Artificial Analysis | Inkling Small HLE | 33.3% |
| Artificial Analysis | Inkling Small AA-LCR | 75.7% |
| Artificial Analysis | Inkling Small τ³-Banking | 18.8% |
| Artificial Analysis | Inkling Small GDPval-AA | 31.6% |
| Artificial Analysis | Inkling Small CritPt | 8.3% |
| Artificial Analysis | Inkling Small HLE-Diamond | 12.1% |
| Artificial Analysis | Inkling Small SciCode | 49.7% |
| Artificial Analysis | Inkling Small Terminal-Bench 4.0 | 1.0% |
| Artificial Analysis | Inkling Small Terminal-Bench 2.1 | 55.1% |
| Artificial Analysis | Inkling Small FrontierCode-AA 2.0 | 4.8% |
| Artificial Analysis | Inkling Small FrontierCode 1.1 | 0.6% |
| Artificial Analysis | Inkling Small Terminal-Bench-Science | 0.0% |
| Artificial Analysis | Inkling Small CritPt v2 (Hard Subset) | 18.5% |
| Artificial Analysis | Inkling Small CritPt v2 | 34.0% |
| Artificial Analysis | Inkling Small AA-Omniscience Accuracy | 31.8% |
| Artificial Analysis | Inkling Small AA-Omniscience Non-Hallucination Rate | 39.5% |
| Artificial Analysis | Inkling Small MLCR-AA | 7.8% |
| Artificial Analysis | Inkling Small MMMU-Pro | 74.0% |
| Artificial Analysis | Inkling Small (High) Intelligence Index | 9.6 |
| Artificial Analysis | Inkling Small (High) Finance & Accounting Index | 20.3 |
| Artificial Analysis | Inkling Small (High) Strategy & Ops Index | 15.2 |
| Artificial Analysis | Inkling Small (High) Legal Index | 18.7 |
| Artificial Analysis | Inkling Small (High) Engineering Index | 18.8 |
| Artificial Analysis | Inkling Small (High) Economics Index | 21.0 |
| Artificial Analysis | Inkling Small (High) Openness Index | 33.3 |
| Artificial Analysis | Inkling Small (High) AutomationBench-AA | 4.1% |
| Artificial Analysis | Inkling Small (High) GDP.pdf | 8.0% |
| Artificial Analysis | Inkling Small (High) HLE-Diamond | 8.4% |
| Artificial Analysis | Inkling Small (High) Terminal-Bench 4.0 | 0.5% |
| Artificial Analysis | Inkling Small (High) FrontierCode-AA 2.0 | 1.5% |
| Artificial Analysis | Inkling Small (High) Terminal-Bench-Science | 0.0% |
| Artificial Analysis | Inkling Small (High) CritPt v2 (Hard Subset) | 9.5% |
| Artificial Analysis | Inkling Small (High) CritPt v2 | 24.3% |
| Artificial Analysis | Inkling Small (High) AA-Omniscience Accuracy | 30.5% |
| Artificial Analysis | Inkling Small (High) AA-Omniscience Non-Hallucination Rate | 36.7% |
| Artificial Analysis | Inkling Small (High) MMMU-Pro | 72.1% |
| Vals AI | Inkling Small Code Migration v1.1 | 10.4% |
| Vals AI | Inkling Small CorpFin v2 | 69.6% |
| Vals AI | Inkling Small CyberBench v1.1 | 64.1% |
| Vals AI | Inkling Small EMB v1.1 | 40.8% |
| Vals AI | Inkling Small Finance Agent v2.1 | 44.8% |
| Vals AI | Inkling Small Vals Finance Index | 36.3% |
| Vals AI | Inkling Small GPQA Diamond | 83.6% |
| Vals AI | Inkling Small HLAB-Vals-1.1 | 2.9% |
| Vals AI | Inkling Small IOI | 9.3% |
| Vals AI | Inkling Small LiveCodeBench | 85.9% |
| Vals AI | Inkling Small LegalBench | 83.0% |
| Vals AI | Inkling Small Vals Legal Index | 57.8% |
| Vals AI | Inkling Small Legal Research Bench | 25.5% |
| Vals AI | Inkling Small MedCode | 37.9% |
| Vals AI | Inkling Small MedScribe | 84.1% |
| Vals AI | Inkling Small MMLU Pro | 85.6% |
| Vals AI | Inkling Small MortgageTax | 62.3% |
| Vals AI | Inkling Small ProgramBench | 0.5% |
| Vals AI | Inkling Small ProofBench v1.1 | 6.0% |
| Vals AI | Inkling Small Public Benefits Bench v1.1 | 59.4% |
| Vals AI | Inkling Small SAGE v1 | 32.7% |
| Vals AI | Inkling Small SkillsBench | 33.6% |
| Vals AI | Inkling Small SWE-bench | 82.2% |
| Vals AI | Inkling Small Tax Agent Bench | 15.4% |
| Vals AI | Inkling Small TaxEval v2 | 75.5% |
| Vals AI | Inkling Small Terminal-Bench 4.0 | 0.0% |
| Vals AI | Inkling Small Vals Index | 27.6% |
| Vals AI | Inkling Small Vals Multimodal Index | 50.1% |
| Vals AI | Inkling Small Vibe Code Bench v1.1 | 19.1% |
Apps
Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for.
Activity
Token volume and request traffic to this model over time.
API
Drop-in code to call this model. OpenRouter's API is OpenAI-compatible — most SDKs work by just swapping the base URL. The only thing that changes between models is the model slug below.
Not currently available through EU in-region routing. In-region routing docs
Frequently asked questions
Inkling Small is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. It is positioned as the smaller, more efficient member of the Inkling family and is suited for reasoning, coding, agentic workflows, retrieval-augmented generation, instruction following, and multilingual conversation.
Inkling Small costs $0.45/M input tokens and $1.20/M output tokens, with separate rates for Cache Read at $0.10/M tokens.
Inkling Small has a 524,288 token context window. It supports up to 262,144 completion tokens.
Yes. Inkling Small accepts tools and tool_choice for function calling. It does not support response_format, so JSON output is not enforced.
Inkling Small accepts text, images, and audio as input and returns text.
Inkling Small is served by 1 provider on OpenRouter: DeepInfra.
Inkling 2 Small and Inkling are other text models from Thinking Machines.
Inkling Small was released on July 30, 2026.