thinkingmachines
Thinking Machines: Inkling 2 Small
thinkingmachines/inkling-2-small
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
thinkingmachines/inkling-2-small is presented by Thinking Machines as an open-weight multimodal mixture-of-experts reasoning model that accepts text and images and produces text. The documented architecture has 594B total parameters with 17B active and supports contexts up to 1M tokens. OpenRouter lists mandatory reasoning with selectable effort plus tool and structured-output controls. Vendor evaluations are supplied only as excerpts, so suitability should be verified against target workloads.[1][2]
Strengths
- Processes image and text inputs together and returns UTF-8 text; the model card describes English as primary with general multilingual ability.[2]
- Uses a sparse MoE decoder with 594B total and 17B active parameters, routing each token to eight of 384 experts.[2]
- Advertises a context window of up to 1M tokens, plus API controls for tools, structured outputs, and several reasoning-effort levels.[1][2]
Limitations
- The OpenRouter catalog marks reasoning as mandatory, so that configuration cannot be fully disabled even though effort levels are adjustable.[1]
- Self-hosting requires substantial capacity: at least 500GB aggregate GPU VRAM for the recommended NVFP4 checkpoint or 800GB for MXFP8.[2]
- Evidence is limited to catalog metadata and vendor documentation or evaluation excerpts; it does not establish behavior for a particular workload.[1][2]
Where to use it
- Consider for assistants that combine screenshots or other images with instructions, after testing visual-task quality on representative inputs.[2]
- Potentially suited to agentic workflows, tool use, coding assistance, and retrieval-augmented generation, all identified by the vendor as intended uses.[1][2]
- May fit long-document or multimodal research applications if the advertised long context performs adequately in workload-specific trials.[1][2]
Sources
- [1] OpenRouter catalog · Thinking Machines: Inkling 2 SmallSources checked: Nov 13, 2026
- [2] Model documentation · Thinking Machines: Inkling 2 SmallSources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
Inkling 2 Small is an open-weight multimodal mixture-of-experts reasoning model from Thinking Machines Lab, with 17B active parameters out of 594B total. It is the first model in the second...
Specifications
- Model ID
- thinkingmachines/inkling-2-small
- Canonical ID
- thinkingmachines/inkling-2-small-20261107
- Creator
- thinkingmachines
- Added to OpenRouter
- 09.11.2026
- Context length
- 1,048,576
- Modality
- text+image->text
- Input modalities
- text, image
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $0.300 / out $1
- 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 | BB — next 30% of the cohort | reasoning: mandatory · default high · efforts: max, high, medium, low, minimal · parameters: include_reasoning, reasoning, reasoning_effort |
| Coding | No measurement | — |
| Agents | No measurement | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: mandatory · default high · efforts: max, high, medium, low, minimal · parameters: include_reasoning, reasoning, reasoning_effort |
| Document analysis | No measurement | — |
| Vision | No measurement | input: image |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Artificial Analysis | Intelligence index | 30.6 | 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) |
|---|---|---|---|---|---|---|
| PrimeIntellectfp4 | 1,048,576 | $0.150 | $0.500 | — | — | 100% |
| DeepInfrafp4 | 262,144 | $0.300 | $1 | — | — | 100% |
| Modalnvfp4 | 1,048,576 | $0.300 | $1 | — | — | 100% |
| Nebiusfp4 | 1,048,576 | $0.300 | $1 | — | — | 100% |
| Together | 1,048,576 | $0.300 | $1 | — | — | 100% |
| Cloudflare | 1,048,576 | $0.300 | $1 | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
No published articles identify this model yet.