Skip to content
RankShark

Inkling

Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total.

ModelOpen sourceby Thinking Machinessince 2026
44.8

SharkScore

#71 in Models

Why it scores 44.8

Inkling has a SharkScore of 44.8/100, ranked #71 of 607 in Models (updated 2026-11-19).

Each part compares Inkling with other listings in Models. Evidence covers 87% of the score's weight.

Adoption
36 · 30%
User reviews
no data · 25%
Maintenance
29 · 20%
Trust & transparency
80 · 15%
Momentum
no data · 10%

Paying never changes this score. How ranking works · method v2026-11-18.2

RankShark review of Inkling

Automated evaluation by RankShark · 18 Nov 2026 · method v1.0 · How we review

We were not paid for this review.

As of 2026-11-18, Inkling is an open-weight text, image, and audio model; OpenRouter lists a 524,288-token context and prices of $1 per million input tokens and $4.05 per million output tokens. [E1][E2]

Inkling is a general-purpose multimodal model from Thinking Machines Lab. It accepts text, image, and audio inputs and generates text. The vendor says it has 975B total parameters and 41B active parameters, and releases open weights for research, fine-tuning, and integration. [E1]

Best for

  • Developers building multimodal applications that process text, images, or audio and return text. · model card
  • Developers building coding assistants, chatbots, or agent and tool-use systems. · model card
  • Researchers and developers seeking open weights for fine-tuning or local deployment. · model card

Not for

  • Teams planning to use outputs for medical, legal, or safety-critical decisions without additional fine-tuning, domain-specific validation, and human oversight. · model card

Key features

  • Multimodal input. The vendor says Inkling accepts text, image, and audio inputs and generates text. [E1] · model card
  • Open weights. The vendor says the weights are released for research, fine-tuning, and integration into third-party products; the model page lists the Apache-2.0 license. [E1] · model card
  • Model architecture. The vendor describes a 975B-parameter model with 41B active parameters and a sparse Mixture-of-Experts backbone. [E1] · model card
  • Local deployment. The vendor lists SGLang, vLLM, TokenSpeed, Unsloth, and Hugging Face recipes for local deployment. [E1] · model card
  • Context and token pricing. OpenRouter lists a 524,288-token context. RankShark measured OpenRouter pricing at $1 per million input tokens and $4.05 per million output tokens on 2026-11-18. [E2] · OpenRouter

Pros

  • The vendor offers open weights and lists the Apache-2.0 license. [E1] · model card
  • It accepts three input modalities: text, image, and audio. [E1] · model card
  • The vendor lists recipes for local deployment with five libraries or frameworks. [E1] · model card

Cons

  • The vendor notes possible hallucinations, instruction-following failures, and degraded performance in long multi-turn conversations. [E1] · model card
  • The vendor reports occasional compliance with role-play or indirectly framed harmful prompts and recommends adding application-layer safeguards. [E1] · model card
  • The vendor says performance may vary across languages, dialects, and less-represented subject areas. [E1] · model card
  • No public prices found on the pages we checked.
  • No free plan or free trial stated on the pages we checked.
  • Whether customer data is used for model training is not stated on the pages we checked.

Pricing

No public prices found on the pages we checked.

Checked 18 Nov 2026

Free plan
Not stated (checked 18 Nov 2026)
Free trial
Not stated (checked 18 Nov 2026)
Trains on customer data
Not stated (checked 18 Nov 2026)
Data retention
Not stated (checked 18 Nov 2026)
Data regions
Not stated (checked 18 Nov 2026)
Data processing agreement
Not stated (checked 18 Nov 2026)
AI models used
Inkling is a multimodal autoregressive transformer with a sparse Mixture-of-Experts backbone. · model card
API
The vendor says API access is available through third-party inference providers and points to the Tinker Cookbook. OpenRouter lists chat completions, completions, messages, and responses endpoints. · model card, OpenRouter
SDKs
Not stated (checked 18 Nov 2026)
Webhooks
Not stated (checked 18 Nov 2026)
Platforms
The vendor lists local deployment recipes for SGLang, vLLM, TokenSpeed, Unsloth, and Hugging Face. · model card
Security claims
No security certification stated (checked 18 Nov 2026)

What we measured

Context window (OpenRouter listing) · now
524,288 tokens
Price per million tokens (OpenRouter listing) · now
$1 in, $4.05 out
Package downloads · 30 days
1,450,118
How we tested · sources
  • Hourly uptime and response-time checks of the product's URL.
  • Repository, release and download data from GitHub, npm, PyPI and Hugging Face where they exist.
  • We did not sign up, pay, or use the product as a customer.
  • 18 Nov 2026: Plans or prices changed on the pricing page.

Something wrong? Report an error with a public link that shows it; corrections re-run the review.

Top alternatives to Inkling

See all
  • 67.1

    Qwen3.8 27B is an open-weight dense vision-language model from Qwen.

    ModelOpen source
  • 65.5

    Qwen3.6-35B-A3B is an open-weight multimodal model from Alibaba Cloud with 35 billion total parameters and 3 billion active parameters per t

    ModelOpen source
  • 64.7

    Qwen3.8 Flash is a multimodal reasoning model from Alibaba.

    ModelOpen source
  • 64.3

    Qwen3.6 27B is a dense 27-billion-parameter language model from the Qwen Team at Alibaba, released in April 2026.

    ModelOpen source