Thinking Machines' Inkling model is now accessible via Vercel AI Gateway, callable with a single identifier: thinkingmachines/inkling in the AI SDK.
Inkling is a broad generalist trained across seven domains simultaneously: agentic tasks, reasoning, coding, instruction-following, factuality, vision, and audio. It supports controllable thinking effort, meaning developers can tune how hard the model works per request. That flexibility, combined with multi-domain training, is the architectural bet worth understanding before you dismiss it as another all-in-one claim.
AI Gateway passes through provider pricing with no markup and no platform fee, including on Bring Your Own Key requests. It adds unified API access, usage tracking, routing rules, Zero Data Retention support, and API key budgets on top. The full changelog details how failover and performance optimization are configured, which is where the operational specifics live.
[READ ORIGINAL →]