# `TypeSafeAPI.Model`
[🔗](https://github.com/typesend/typesafe_ai/blob/v0.1.0-alpha.3/lib/typesafe_api/model.ex#L1)

A model available to the account, as listed by `TypeSafeAPI.models/1`.

`release_date` is a `Date`, because that is what the endpoint means: the
OpenAPI schema describes it as "Model release date, formatted as YYYY-MM-DD",
even though the live API sends a full ISO 8601 datetime today. A datetime is
reduced to the calendar date it carries, without shifting it into UTC, so a
model released on the 15th at 20:00-07:00 does not report the 16th. A value
this library cannot parse as a date at all — the `"latest"` alias, say —
leaves `release_date` `nil`.

`release_date_raw` is the string exactly as it arrived, so nothing is lost to
the normalization. `raw` is the whole entry, so a field this struct does not
model yet (a context window, a deprecation date) is reachable without waiting
for a release; this endpoint is reverse-engineered from the official Python
SDK, so it is the most likely one to grow fields.

`description` is required and non-nullable in the spec, but a non-string one
degrades to `nil` here rather than failing the call. It stays in `raw`.

# `t`

```elixir
@type t() :: %TypeSafeAPI.Model{
  description: String.t() | nil,
  name: String.t(),
  raw: map(),
  release_date: Date.t() | nil,
  release_date_raw: String.t() | nil
}
```

# `decode`

```elixir
@spec decode(term()) :: {:ok, t()} | :error
```

Decodes one entry of the `models` array.

Only a string `name` is required; `TypeSafeAPI.Models.list/2` skips an entry
that returns `:error` rather than failing the whole list.

---

*Consult [api-reference.md](api-reference.md) for complete listing*