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

# Platform Status & Roadmap

> Verified current behavior, known gaps, and the work required to close them.

## Verified current behavior

| Area                | Current evidence                                                                                                                                            |
| ------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Client encryption   | Direct SDK and local proxy requests completed through the encrypted RVENC path                                                                              |
| Chat                | Live non-streaming and streaming requests completed on OpenAI and Anthropic compatibility surfaces                                                          |
| Coding clients      | OpenCode `1.18.11`, Claude Code `2.1.220`, OpenClaw `2026.7.1-2`, Hermes Agent `0.20.0`, and Goose `1.45.0` completed live requests through the local proxy |
| Audio transcription | `deepgram/general-nova-3` completed a live request and returned Deepgram-style results                                                                      |
| Model catalogue     | The current documented aliases are listed below                                                                                                             |
| Attestation gate    | The SDK calls high-level Reticle attestation and requires a session ID before protected requests when attestation is enabled                                |

## Attestation status

| Capability                          | Status                                | Boundary                                                                                                                                                       |
| ----------------------------------- | ------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| AMD SEV-SNP evidence verification   | Implemented                           | Signature/certificate and nonce checks run; expected measurement comparison remains TODO                                                                       |
| Intel TDX evidence verification     | Implemented                           | Quote collateral and nonce verification run; expected measurement comparison remains TODO                                                                      |
| NVIDIA evidence verification        | Implemented with source-review caveat | Overall and detached JWTs, digest links, nonce values, and returned GPU claims are processed; selected per-GPU check semantics need code-owner review          |
| Combined module discovery           | Implemented                           | The unified client discovers one CPU module and an optional NVIDIA evidence bundle; the bundle can contain multiple detached GPU claims                        |
| Automatic approved-image comparison | Not implemented in reviewed CPU paths | Required before claiming that the verifier proves the expected release image                                                                                   |
| Multi-GPU serving-path appraisal    | Partial                               | The token parser handles multiple returned GPU claims, but the public result does not expose per-GPU verdicts or prove exact-set session and scheduler binding |

Attestation evidence can still provide meaningful hardware authenticity and freshness checks. The missing policy checks limit the conclusion: a pass must not be described as complete proof that an approved release image processed the request.

## Model status

The documented model set includes:

* `qwen36-27b`;
* `glm-5.2`, resolving to `zai-org/GLM-5.2`;
* `deepgram/general-nova-3` for audio transcription.

## Known gaps

| Gap                         | Impact                                                                                                                                |
| --------------------------- | ------------------------------------------------------------------------------------------------------------------------------------- |
| Multi-GPU evidence contract | Developers cannot audit from the public result that every participating GPU, and only those GPUs, were appraised and bound to serving |
| Reproducible enclave images | A verifier cannot independently rebuild a release and connect the resulting measurements to production                                |

## Priority sequence

1. Publish reproducible images, provenance, and release-to-measurement reference sets.
2. Expose per-GPU verdicts, review the current check semantics, and enforce exact-set backend binding for multi-GPU inference.
3. Resolve package/runtime version reporting.

## Trust progression

```mermaid theme={"system"}
flowchart LR
    A["Manufacturer evidence<br/>signature, collateral, nonce"] --> B["Release policy<br/>approved measurements and TCB"]
    B --> C["Route assurance<br/>model and backend classification"]
    C --> D["Reproducible provenance<br/>source to deployed measurement"]
```

Each stage answers a different question. Manufacturer evidence authenticates hardware claims. Release policy decides whether measured values are approved. Route assurance binds that decision to a model path. Reproducible provenance connects the approved measurements to source and build inputs.

<Note>
  Status entries distinguish source review from live testing. The presence of an API route alone does not establish that a model or end-to-end workflow is operational.
</Note>