Thinking Machines / thinkingmachines/inkling
Inkling - access through LLMTR
Inkling is Thinking Machines' flagship open-weights model: a Mixture-of-Experts architecture with 975 billion total and 41 billion active parameters per token. It reasons step by step before answering, and you can tune that effort per request — six levels from none (off) to xhigh via `reasoning_effort`. It accepts images as base64 data URLs, makes tool/function calls, produces JSON object output, and supports prompt-cache reads. Reasoning tokens are returned inside `completion_tokens` and billed as output.
Technical specifications
| Canonical ID | thinkingmachines/inkling |
|---|---|
| Provider | Thinking Machines |
| Context window | 65,536 tokens |
| Operations | CHAT_COMPLETIONS |
| Modalities | text, image |
Pricing
A 6% platform margin applies to credit top-ups; model usage prices are not separately marked up.
| Operation | Metric | Unit | Price |
|---|---|---|---|
| CHAT_COMPLETIONS | INPUT_TEXT | PER_1M_TOKENS | $1.87 |
| CHAT_COMPLETIONS | CACHE_READ | PER_1M_TOKENS | $0.374000 |
| CHAT_COMPLETIONS | OUTPUT_TEXT | PER_1M_TOKENS | $4.68 |
Example usage
With existing OpenAI SDK flows, change only the base URL and model identifier.
curl https://llmtr.com/v1/chat/completions \
-H "Authorization: Bearer llmtr-your_key" \
-H "Content-Type: application/json" \
-d '{"model":"thinkingmachines/inkling","messages":[{"role":"user","content":"Hello"}]}'Related models
- Gemma 4 llmtr/gemma-4
- Qwen 3.6 35B-A3B llmtr/qwen3-6-35b
- MedGemma 4B llmtr/medgemma-4b
- Trendyol 7B llmtr/trendyol-7b
- Magibu 11B v8 llmtr/magibu-11b-v8
- Sincap llmtr/sincap
- EmbeddingGemma 300M llmtr/embeddinggemma-300m
- Qwen 3.5 4B llmtr/qwen3-5-4b