thinkingmachines
Thinking Machines: Inkling
thinkingmachines/inkling
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
Inkling is a general-purpose open-weight multimodal model from Thinking Machines Lab. The model documentation describes text, image and audio inputs with UTF-8 text output, a 975B-parameter sparse mixture-of-experts architecture with 41B active parameters, and support for English plus general multilingual use. OpenRouter reports a 524,288-token context window and configurable reasoning effort. Its documentation positions it for coding, agentic and tool-use systems, chatbots and retrieval-augmented generation; reported benchmark and safety results should be treated as vendor-reported evidence rather than independent validation.[1][2]
Strengths
- Accepts text, image and audio inputs and produces UTF-8 text; the documentation describes a natively multimodal sparse MoE transformer.[2]
- Open weights, BF16/NVFP4 support and published local-serving recipes support research, fine-tuning and deployment outside a hosted API.[2]
- OpenRouter reports tool calling, configurable reasoning effort and a 524,288-token context window for long-context or agentic workloads.[1]
Limitations
- The evidence is limited to OpenRouter catalog data and the vendor's model documentation; no independent evaluation is supplied, so validate on real workloads.[1][2]
- The documentation warns of hallucination, imperfect instruction following, degraded long multi-turn behavior, uneven language or domain performance, and residual compliance with harmful role-play or indirectly framed prompts.[2]
- For best results, the documentation gives practical input bounds: images ideally 40-4096 pixels per side and 16 kHz WAV audio ideally under 20 minutes.[2]
Where to use it
- A reasonable candidate for assistants that answer text questions using image or audio context, subject to multimodal validation.[1][2]
- Potentially suitable for coding assistants, agentic tool-use systems, chatbots and RAG, with application-level evaluation and safeguards.[1][2]
- Worth considering when open weights, fine-tuning or local deployment are project requirements.[2]
Sources
- [1] OpenRouter catalog · Thinking Machines: InklingSources checked: Nov 13, 2026
- [2] Model documentation · Thinking Machines: InklingSources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total. It is designed for general-purpose reasoning, coding, agentic and tool-use systems,...
Specifications
- Model ID
- thinkingmachines/inkling
- Canonical ID
- thinkingmachines/inkling-20260715
- Creator
- thinkingmachines
- Added to OpenRouter
- 17.07.2026
- Context length
- 524,288
- Modality
- text+image+audio->text
- Input modalities
- text, image, audio
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $1 / out $4.05
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | CC — below the B band | reasoning: optional · default high · efforts: max, high, medium, low, minimal, none · parameters: include_reasoning, reasoning, reasoning_effort |
| Coding | BB — next 30% of the cohort | — |
| Agents | BB — next 30% of the cohort | parameters: tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: optional · default high · efforts: max, high, medium, low, minimal, none · parameters: include_reasoning, reasoning, reasoning_effort |
| Document analysis | No measurement | — |
| Vision | No measurement | input: image |
| Image generation | No measurement | — |
| Audio | No measurement | input: audio |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Artificial Analysis | Intelligence index | 12.5 | CC — below the B band | 13.11.2026 |
| Artificial Analysis | Coding index | 52.1 | BB — next 30% of the cohort | 13.11.2026 |
| Artificial Analysis | Agentic index | 22.5 | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | agents · webappswin rate 48.6% · rank 28 | 1,177Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · 3dwin rate 39.9% · rank 71 | 1,145Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · asciiartwin rate 34.1% · rank 67 | 1,087Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · codecategorieswin rate 41.9% · rank 66 | 1,195Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · datavizwin rate 37.8% · rank 79 | 1,158Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · gamedevwin rate 37.4% · rank 75 | 1,153Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · svgwin rate 34.5% · rank 65 | 1,102Elo | CC — below the B band | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 40.1% · rank 68 | 1,190Elo | BB — next 30% of the cohort | 13.11.2026 |
| Design Arena | models · websitewin rate 44.4% · rank 61 | 1,216Elo | BB — next 30% of the cohort | 13.11.2026 |
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| Together | 524,288 | $1 | $4.05 | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.