SYSTEM Cited by 1 source
Tinker (Thinking Machines)¶
Tinker (thinkingmachines.ai/tinker) is Thinking Machines' LLM fine-tuning product. First canonical wiki reference: sources/2026-02-13-netflix-scaling-llm-post-training-at-netflix — where Netflix explicitly contrasts Tinker against its internal Post-Training Framework to justify building in-house.
Netflix's stated contrast¶
"Existing tools (e.g., Thinking Machines' Tinker) work well for standard chat and instruction-tuning, but their structure can limit deeper experimentation. In contrast, our internal use cases often require architectural variation (for example, customizing output projection heads for task-specific objectives), expanded or nonstandard vocabularies driven by semantic IDs or special tokens, and even transformer models pre-trained from scratch on domain-specific, non-natural-language sequences. Supporting this range requires a framework that prioritizes flexibility and extensibility over a fixed fine-tuning paradigm."
Position in the landscape¶
Tinker represents the "standardised fine-tuning product" end of the LLM post-training toolchain — opinionated, ergonomic, optimised for common chat/instruction-tuning shapes. Netflix's framework sits at the opposite end: more complexity surfaced to the user in exchange for architectural freedom (custom output heads, non-standard vocabularies, non-NL transformer architectures).
This is the same product-vs-platform trade-off that recurs throughout ML infrastructure design — packaged workflows optimise for speed-to-first-trained-model but constrain teams whose use cases fall outside the canonical shape.