Flint review
Flint is out. Launch facts from official sources are below, and our hands-on review follows.
Last checked 2 min ago.
Launch facts
| Item | Value | Source |
|---|---|---|
| Released | Oct 27, 2026 | First seen live by Modelcast |
| Context size (tokens) | 262,144 | openrouter.ai |
| Official download | https://huggingface.co/RecursiveSI/flint | huggingface.co |
| License | apache-2.0 | huggingface.co |
| Longest answer (tokens) | 235,929 | openrouter.ai |
| Reads | text | openrouter.ai |
| Size (parameters) | 39.8B | huggingface.co |
We only show a value when an official source states it, and we keep the page and the exact words it came from. Anything we cannot prove says so.
What we predicted vs what shipped
| Item | We said | What shipped | Result |
|---|---|---|---|
| Launch date | 2026-11-04 (window 2026-10-27 to 2026-11-20, 60% chance) | Oct 27, 20268 days earlier than our most likely date | Close |
| Memory | Reads up to 262,144 tokens at once (about 200,000 words) | 262144 | Close |
| Download | Docs point to RecursiveSI/flint on Hugging Face. Not yet checked | https://huggingface.co/RecursiveSI/flint | Nothing to compare |
Our forecast is kept exactly as it was published, so you can check our record. See the track record.
How it got here
| Date | What happened | Source type |
|---|---|---|
| Oct 27, 2026 | Flint is out. | Confirmed |
| Oct 21, 2026 | Flint docs and config added to Transformers main, credited to Recursive.Hugging Face Transformers, main branch vs PyPI 5.20.2 (checked 26 Oct 2026) | Leak |
| Oct 19, 2026 | Transformers 5.20.2 is the latest release and has no Flint support.Hugging Face Transformers, main branch vs PyPI 5.20.2 (checked 26 Oct 2026) | Confirmed |
Frequently asked questions
When was Flint released?
Oct 27, 2026.
Is Flint worth using?
Our review is in testing. The launch facts are above.
Where can I download or use Flint?
See the official Recursive site.
Is there a full review?
It goes up on this page once testing is done.
Related models
Kolibri
Aleph AlphaRuns cheaply on-premise, which companies with strict data rules need.
Nemotron 3.5 Lightning
NVIDIAOnly 3B of its 30B parameters work at a time, so it is one of the fastest capable models on a single GPU.
Muse Glimmer
MetaMeta's only free model this year. Many people searching for Llama 5 actually want this.