> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tinfoil.sh/llms.txt
> Use this file to discover all available pages before exploring further.

# Decision models

> Decision models available on Tinfoil for classification and rubric scoring with calibrated probabilities.

> [!WARNING]
>
> Decisions need recent SDK versions: Python `tinfoil >= 0.15.0`, Go
> `tinfoil-go >= v0.15.10`, JavaScript `tinfoil >= 1.3.0`. Swift and Rust do
> not support decisions yet.

Decision models answer typed questions about an input and return calibrated
probabilities instead of generated text. See the [Decision Outputs guide](https://docs.tinfoil.sh/guides/decision-outputs.md)
for more.

> [!NOTE]
>
> Decision models are a rapidly changing area right now, hence the generic `decision-small`, `decision-medium`, and
> `decision-large`, reflecting their relative size tiers. Today all three are
> served by the [vllm-sr](https://huggingface.co/vllm-sr) d3 family, which perform well on the
> [Jev Decision Index](https://huggingface.co/spaces/multimodalart/jev-decision-index).
> As better decision models appear, the checkpoints behind these names may
> change.

> [card]
>
> ![Decision models](https://mintcdn.com/tinfoil/7N4r3KvDOY8x2u0o/images/model-icons/decision.png?fit=max\&auto=format\&n=7N4r3KvDOY8x2u0o\&q=85\&s=6bb6dd59d44de03db82beec5ef24e7e0)
>
> Decision Large
>
> [`decision-large`](#decision-large)
>
> **Parameters:** 27B
>
> **Quantization:** None (served in BF16)
>
> **Capabilities:** Typed decisions: true/false probabilities, choices from your options, and rubric scores
>
> **Context Window:** 32k tokens
>
> **Best for:** Tricky decision tasks. The slowest of the family, though still quicker than outputting text.
>
> **Model weights:** [vllm-sr/d3](https://huggingface.co/vllm-sr/d3)
>
> **Configuration repo:** [tinfoilsh/confidential-decision-models](https://github.com/tinfoilsh/confidential-decision-models)

> [card]
>
> ![Decision models](https://mintcdn.com/tinfoil/7N4r3KvDOY8x2u0o/images/model-icons/decision.png?fit=max\&auto=format\&n=7N4r3KvDOY8x2u0o\&q=85\&s=6bb6dd59d44de03db82beec5ef24e7e0)
>
> Decision Medium
>
> [`decision-medium`](#decision-medium)
>
> **Parameters:** 9B
>
> **Quantization:** None (served in BF16)
>
> **Capabilities:** Typed decisions: true/false probabilities, choices from your options, and rubric scores
>
> **Context Window:** 32k tokens
>
> **Best for:** Most decision tasks: escalation, policy evaluation, triage. The default choice.
>
> **Model weights:** [vllm-sr/d3-flash](https://huggingface.co/vllm-sr/d3-flash)
>
> **Configuration repo:** [tinfoilsh/confidential-decision-models](https://github.com/tinfoilsh/confidential-decision-models)

> [card]
>
> ![Decision models](https://mintcdn.com/tinfoil/7N4r3KvDOY8x2u0o/images/model-icons/decision.png?fit=max\&auto=format\&n=7N4r3KvDOY8x2u0o\&q=85\&s=6bb6dd59d44de03db82beec5ef24e7e0)
>
> Decision Small
>
> [`decision-small`](#decision-small)
>
> **Parameters:** 4B
>
> **Quantization:** None (served in BF16)
>
> **Capabilities:** Typed decisions: true/false probabilities, choices from your options, and rubric scores
>
> **Context Window:** 32k tokens
>
> **Best for:** Very fast, high-volume decisions, with some accuracy tradeoff. Works well as a first pass.
>
> **Model weights:** [vllm-sr/d3-mini](https://huggingface.co/vllm-sr/d3-mini)
>
> **Configuration repo:** [tinfoilsh/confidential-decision-models](https://github.com/tinfoilsh/confidential-decision-models)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.