Tool overview
Inkling is listed under AI Models & Platforms AI tools.
What is Inkling?
Inkling is a general-purpose multimodal model from Thinking Machines Lab. It accepts text, image, and audio inputs, produces text, supports a one-million-token context window, and is distributed through open weights, Tinker API access, and third-party inference providers.
Best for
Developers and researchers needing an open multimodal foundation model
Who is it for?
Decision note
Suitable for evaluation after confirming final commercial terms, permissions, data handling, lifecycle status, and edit-screen aliases. Keep Needs Review enabled until a human verifies the published profile.
Key features
Text, image, and audio input with text output
975B total parameters with 41B active
Context window up to one million tokens
Open weights under Apache 2.0
Tinker and third-party API access
Support for agentic coding and tool-use workflows
Use cases
Building coding assistants
Creating multimodal chatbots
Running tool-using AI agents
Fine-tuning domain applications
Deploying self-hosted research models
Pros
- Apache-2.0 open weights support broad integration
- Handles text, image, and audio inputs
- Available through both self-hosted and API paths
Cons
- Self-hosting requires very large GPU memory
- Open weights do not remove safety and governance needs
- Hosted inference and fine-tuning costs depend on providers
Limitations
Inkling can hallucinate, miss instructions, reflect training-data bias, and degrade in long interactions. Self-hosting requires at least 600 GB aggregated VRAM for the NVFP4 checkpoint or substantially more for BF16, plus application-level safeguards and evaluation.
Pricing details
Billing options
Pricing note
Inkling weights are publicly downloadable under Apache 2.0, so the model is recorded as Free with an ongoing Free Plan and no time-limited Free Trial. Hosted Tinker or third-party inference and fine-tuning can incur separate provider charges.
Supported languages
- English
- General multilingual support
Integrations
Hugging Face Transformers
vLLM
SGLang
Tinker
Third-party inference providers
Docker Model Runner




Please log in to join the discussion.