The Orchestrator › Models › Thinking Machines

Thinking Machines: Inkling

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

ProviderThinking Machines
Model IDthinkingmachines/inkling
Context window524K tokens (524,288)
Modalitiestext + image + audio → text
TokenizerOther
Router sourcesOpenRouter, TrustedRouter, Vercel AI Gateway, Cloudflare
Released2026-07-17
Tracked since2026-11-14
PopularityNot in the ranked set
ReasoningOn by default

Pricing (per million tokens)

VariantSourceAPI IDInputOutputContext
standardOpenRouterthinkingmachines/inkling$1.00$4.05524K
standardTrustedRouterthinkingmachines/inkling$1.06$4.27524K
standardVercel AI Gatewaythinkingmachines/inkling$1.00$4.05256K
standardCloudflarethinkingmachines/inkling$1.87$4.68—

Recorded price history

DateInput / 1MOutput / 1MContext
2026-11-14$1.87$4.68—
2026-11-14$1.00$4.05256K
2026-11-14$1.06$4.27524K
2026-11-14$1.00$4.05524K

Capabilities

Vision (image input)Audio inputAgentic tool useReasoning mode256K+ context windowControllable reasoningNo content storage (attested)

Trust flags come from TrustedRouter's per-provider attestations.

Supported parameters

frequency_penalty, include_reasoning, logit_bias, max_tokens, min_p, presence_penalty, reasoning, reasoning_effort, repetition_penalty, stop, temperature, tool_choice, tools, top_k, top_p