Sovereign inference — Inkling 2 Small

Your sovereign, private Inkling 2 Small

Inkling 2 Small is Thinking Machines Lab's open-weight (Apache 2.0) reasoning model: a 594B-parameter mixture-of-experts with 17B active parameters, a 1M-token context window and controllably interleaved thinking. It accepts text and image input and is built for long-horizon agentic work, tool orchestration and code generation, sustaining multi-turn sessions on large document sets. A reasoning model in the medium tier, served on ToothFairyAI's AU, EU and US endpoints — private by default, billed in US dollars per million tokens at the published rates below. No training on your data, no vendor lock-in.

AU · EU · US resident$0.42/1M input$1.51/1M output$0.04/1M cached input1049K context

How Inkling 2 Small compares

Independent intelligence index for Inkling 2 Small against every peer in the Medium tier (and the tier above where needed) — same source, same tests.

Inkling 2 Small (this model)30.6DeepSeek V4.1 Flash 1017 Ultra-fast45.1GLM 5.3 Flash Priority41.8DeepSeek V4.1 Flash Priority39.5DeepSeek V4 Flash Official34.3Minimax 329.2GPT OSS 20B9.0

Throughput — tokens per second, median

Inkling 2 Small117.6 t/sDeepSeek V4.1 Flash 1017 Ultra-fast210.6 t/s

Latency — end-to-end seconds for a 500-token answer

DeepSeek V4.1 Flash 1017 Ultra-fast13.0sGLM 5.3 Flash Priority58.6sDeepSeek V4.1 Flash Priority11.7sDeepSeek V4 Flash Official12.5sMinimax 322.5sGPT OSS 20B14.7s

Capability indexes

Domain-weighted agentic performance for Inkling 2 Small.

Finance & AccountingStrategy & OpsLegalHealthcare & MedicalEngineeringEconomics

Capability index vs medium peers

13.126.239.452.544.343.541.640.838.943.4Inkling 2 Small48.352.546.445.250.5DeepSeek V4.1 Flash 1017 Ultra-fast42.045.139.944.843.848.9GLM 5.3 Flash Priority45.051.741.340.639.344.4DeepSeek V4.1 Flash Priority37.543.137.735.434.940.3DeepSeek V4 Flash Official30.125.031.529.731.243.7Minimax 37.34.87.36.611.711.3GPT OSS 20B
Finance & AccountingStrategy & OpsLegalHealthcare & MedicalEngineeringEconomics

The evaluations behind the index

Independent benchmark results for Inkling 2 Small — Elo scores as published; pass rates shown as percentages.

AutomationBench-AA45.6 %Terminal-Bench 4.013.1 %SciCode52.3 %Humanity's Last Exam44.1 %GDP.pdf25.2 %AA-LCR (long context)83.3 %

Per-evaluation economics

EvaluationScoreTime / taskOutput tokens / task
AA-Briefcase1526.8623.0 min162347
GDPval-AA1630.2313.5 min95492
AutomationBench-AA45.6%2.7 min18917
Terminal-Bench 4.013.1%41.1 min289846
SciCode52.3%1.4 min9825
Humanity's Last Exam44.1%5.6 min39616
GDP.pdf25.2%3.4 min23984
AA-LCR (long context)83.3%0.7 min4612

Run Inkling 2 Small privately

One key, one endpoint, three jurisdictions.

  • Size tierMedium
  • Serving routeStandard
  • Input rate$0.42/1M
  • Output rate$1.51/1M
  • Cached input rate$0.04/1M
  • Context window1049K tokens
  • Max output262K tokens
  • ReasoningYes — emits thinking traces
  • Vision & video inputSupports image input
  • Tool callingSupported
  • Data residencyAU, EU and US endpoints — processing stays inside the region you call