Fireship · 2026-07-20 · notable
Fireship: 'This $12 billion startup finally shipped something...'
Fireship reviews Inkling, the 975B-parameter open-weights model from Mira Murati's Thinking Machines Lab, five days after its release. He calls the model 'deliberately mid' and walks through why it still matters.
Fireship's 21-minute breakdown of Thinking Machines Lab's first open-weights release, Inkling.
What is it?
The Fireship video reviews Inkling, the debut open-weights model from Mira Murati's Thinking Machines Lab. Thinking Machines closed a $2B seed round at a $12B valuation last year and had not shipped a public model until Inkling landed on Hugging Face on July 15. Fireship frames the video around what a $12B pre-product startup finally delivered.
How does it work?
Fireship walks through Inkling's specs — a 975B-parameter sparse Mixture-of-Experts transformer with 41B active parameters, trained on 45T multimodal tokens across text, images, audio, and video — and its Apache-2.0 license, controllable thinking effort, and 1M-token context window. He labels the model 'deliberately mid' relative to top closed models, arguing Thinking Machines is aiming at a fine-tuning-friendly base for customization rather than a leaderboard SOTA.
Why does it matter?
Thinking Machines was one of the highest-profile AI bets of the last two years, funded partly on Murati's OpenAI pedigree. Inkling gives the outside world its first concrete artifact to judge. Fireship's take — the model is real, the strategy is niche, the valuation is a stretch — is a useful signal for developers deciding whether to build on Inkling via Tinker or stay on established open weights like Qwen and DeepSeek.
Who is it for?
developers weighing open-weights options and anyone tracking Thinking Machines Lab
Try it
Watch: youtube.com/watch?v=M51asSwRLxA