Remove sunset GitHub Models integration

CHANGELOG.md

@@ -4,6 +4,19 @@ All notable changes to this project will be documented in this file.
 
 ## [Unreleased]
 
+### Fixed
+- **`sonnet` now maps to Sonnet, not Opus.** The default model mappings folded
+  every Claude spelling — Sonnet included — into the newest Opus. That is
+  right for aliases actually named after a Claude version Copilot dropped, but
+  `sonnet` and every `claude-sonnet-*` spelling now resolve to `claude-sonnet-5`
+  instead, so a caller that asked for the mid tier gets its cost and rate
+  limits rather than being silently upgraded to Opus.
+
+### Removed
+- Removed GitHub Models routing, catalog merging, token configuration, and setup
+  now that the GitHub Models inference service has been retired. All model
+  requests now use GitHub Copilot.
+
 ## [1.4.3] - 2026-08-09
 
 ### Added

README.md

@@ -49,11 +49,6 @@ file.
   as `response_format`. `topK` and `safetySettings` have no counterpart and are
   dropped, which the dashboard states rather than leaving you to assume they
   took effect.
-- **GitHub Models inference** — requests whose model id uses the
-  `publisher/model` form (e.g. `openai/gpt-4o`) are transparently routed to the
-  [GitHub Models](https://models.github.ai) API instead of Copilot, authenticated
-  with a token that has the `models: read` permission. Enabled by
-  default; the catalog is merged into `/v1/models`.
 - **Optional API-key authentication** on the LLM endpoints
   (`Authorization: Bearer`, `x-api-key`, or `x-goog-api-key`), disabled by
   default and compared in constant time.
@@ -134,11 +129,7 @@ The GitHub token is exchanged for a short-lived **Copilot token** via
 `https://api.github.com/copilot_internal/v2/token`, which is refreshed
 automatically before it expires.
 
-The interactive Device Flow requests the `read:user copilot` scopes. GitHub
-Models is **not** covered by the Device Flow token (`models` is not a valid
-classic OAuth scope). To use the [GitHub Models](#github-models) inference API,
-supply a dedicated token with the `models: read` permission (a fine-grained PAT)
-via `github_models.token`.
+The interactive Device Flow requests the `read:user copilot` scopes.
 
 ## Endpoint Authentication
 
@@ -180,7 +171,7 @@ Config file: `~/.ghc-tunnel/config.yaml` (`%APPDATA%/ghc-tunnel/config.yaml`
 on Windows). It is generated on first run or with `--config`.
 
 ```yaml
-config_version: 4
+config_version: 6
 address: 127.0.0.1
 port: 8314
 debug: false
@@ -192,14 +183,10 @@ auto_upgrade: true                  # self-update on startup; false to disable
 model_mappings:
   exact:
     opus: claude-opus-5
-    sonnet: claude-opus-5
+    sonnet: claude-sonnet-5
     haiku: claude-haiku-4.5
   prefix:
-    claude-sonnet-4-: claude-opus-5
-github_models:
-  enabled: true                     # route publisher/model ids to GitHub Models
-  # org: my-org                     # attribute inference to an organization
-  # token: ghp_xxx                  # dedicated token (models: read permission)
+    claude-sonnet-4-: claude-sonnet-5
 system_prompt_remove: []
 system_prompt_add: []
 tool_result_suffix_remove: []
@@ -211,52 +198,6 @@ upstream_read_timeout_seconds: 900   # max silence from upstream; 0 disables
 # api_key: my-secret-key
 ```
 
-## GitHub Models
-
-Besides Copilot, GitHub offers a separate model **inference** service —
-[GitHub Models](https://models.github.ai) — exposing OpenAI-compatible endpoints
-for models from OpenAI, Meta, Mistral, xAI, DeepSeek, and others. This proxy
-routes to it transparently.
-
-**Routing.** GitHub Models identifies models by a `publisher/model` id (e.g.
-`openai/gpt-4o`, `meta/llama-4-maverick`). When `github_models.enabled` is true
-(the default), any request whose *translated* model id contains a `/` is sent to
-GitHub Models instead of Copilot. Because Copilot model ids never contain a `/`,
-the two never collide, and existing model mappings are unaffected. This works on
-`/v1/chat/completions`, `/v1/messages` (translated), and the Gemini endpoints.
-
-```bash
-curl http://127.0.0.1:8314/v1/chat/completions \
-  -H "Content-Type: application/json" \
-  -d '{"model": "openai/gpt-4o", "messages": [{"role": "user", "content": "Hi!"}]}'
-```
-
-**Authentication.** GitHub Models uses the raw GitHub token (not the Copilot
-token) via `Authorization: Bearer`. The token must carry the **`models: read`**
-permission (a fine-grained PAT). The token minted by the Device Flow does **not**
-have this permission (`models` is not a valid classic OAuth scope), so set a
-dedicated token via `github_models.token` or the
-`GHC_PROXY_GITHUB_MODELS_TOKEN` environment variable. Tokens without the
-permission get an `Unauthorized` response from GitHub.
-
-**Configuration.**
-
-```yaml
-github_models:
-  enabled: true          # set false to always use Copilot
-  org: my-org            # optional: attribute inference to an organization
-  token: ghp_xxx         # optional: dedicated token (models: read permission)
-```
-
-| Environment variable | Effect |
-|----------------------|--------|
-| `GHC_PROXY_GITHUB_MODELS_ENABLED` | Enable/disable routing (`true`/`1`) |
-| `GHC_PROXY_GITHUB_MODELS_ORG` | Attribute inference to an organization |
-| `GHC_PROXY_GITHUB_MODELS_TOKEN` | Dedicated token for GitHub Models |
-
-The GitHub Models catalog is merged into `GET /v1/models` so those ids show up in
-the dashboard and model listings.
-
 ## API Endpoints
 
 | Endpoint | Description |
@@ -271,6 +212,9 @@ the dashboard and model listings.
 | `POST /v1beta/models/{model}:generateContent` | Gemini generate content |
 | `POST /v1beta/models/{model}:streamGenerateContent` | Gemini streaming (SSE) |
 | `POST /v1beta/models/{model}:countTokens` | Gemini token counting |
+| `POST /v1/embeddings` | Embeddings (also `/embeddings`) |
+| `GET /v1/models/full/` | Raw upstream model catalog with capabilities |
+| `GET /usage` | Copilot plan and quota usage (also `check-usage`) |
 | `GET /health` | Liveness/readiness probe (`?strict=true` for 503 when not ready) |
 | `GET /openapi.json` | OpenAPI v3 specification |
 | `GET /` | Web dashboard — overview |
@@ -282,6 +226,9 @@ the dashboard and model listings.
 | `POST /api/config/debug` | Turn body capture on or off (`{"debug": true}`) |
 | `GET /api/stats` | Running totals, including what Copilot billed |
 | `GET /api/cache` | Prompt-cache statistics, overall and per model |
+| `GET /api/requests` | Recent requests (JSON) |
+| `GET /api/audit` | Filtered audit records |
+| `GET /api/audit/summary` | Aggregated audit summary |
 | `GET /api/models` | All supported models (used by the dashboard) |
 
 The `/api/config/` routes are guarded by the API key when one is configured;
@@ -335,6 +282,7 @@ cargo clippy     # lint
 | File | Responsibility |
 |------|----------------|
 | `src/main.rs` | CLI parsing and server startup |
+| `src/lib.rs` | Library crate root re-exporting the modules below |
 | `src/setup.rs` | Interactive first-run setup wizard |
 | `src/config.rs` | Config dir, YAML config, defaults, model-mapping defaults |
 | `src/auth.rs` | GitHub token resolution (env/file/Device Flow), Copilot token exchange |

docs/api.md

@@ -357,11 +357,6 @@ counts, estimated cost, and prompt-cache hit rate.
 
 ## Notable behaviors
 
-- **GitHub Models routing** — when enabled (default), requests whose translated
-  model id uses the `publisher/model` form (e.g. `openai/gpt-4o`) are routed to
-  the [GitHub Models](https://models.github.ai) inference API instead of Copilot,
-  authenticated with a token that has the `models: read` permission.
-  See [Configuration](configuration.md#github-models).
 - **Model translation** — model names are rewritten per your
   [mappings](configuration.md#model-mappings) before being forwarded.
 - **1M context** — for Anthropic-native requests, the proxy forwards the

docs/claude-code.md

@@ -63,9 +63,10 @@ model_mappings:
     claude-opus-4-8: claude-opus-5
 ```
 
-The built-in defaults already do this for every Claude spelling. They apply to
-a *new* config file only — an existing one keeps the targets it has, so if you
-wrote yours before Opus 5 shipped, either edit it or re-run `--setup`.
+The built-in defaults already do this for every Opus and Sonnet spelling. They
+apply to a *new* config file only — an existing one keeps the targets it has,
+so if you wrote yours before Opus 5 shipped, either edit it or re-run
+`--setup`.
 
 Restart the proxy after editing `config.yaml` — mappings are read at startup.
 

docs/configuration.md

@@ -21,7 +21,7 @@ Windows). Generated on first run, with `--config`, or through the setup wizard.
 
 ```yaml
 # Schema version for migration/write-back behavior
-config_version: 4
+config_version: 6
 
 # Server settings
 address: 127.0.0.1
@@ -46,15 +46,7 @@ model_mappings:
     sonnet: claude-sonnet-5
     haiku: claude-haiku-4.5
   prefix:
-    claude-sonnet-4-: claude-opus-5
-
-# GitHub Models (https://models.github.ai) inference
-# Route publisher/model ids (e.g. openai/gpt-4o) to GitHub Models instead of
-# Copilot. Needs a token with the `models: read` permission (fine-grained PAT).
-github_models:
-  enabled: true
-  # org: my-org
-  # token: ghp_xxx
+    claude-sonnet-4-: claude-sonnet-5
 
 # Content filtering
 system_prompt_remove: []
@@ -85,8 +77,9 @@ Incoming model names are rewritten before the request is forwarded upstream:
 Exact matches take priority over prefix matches. Unmapped names pass through
 unchanged. Use the live catalog at `GET /v1/models` to discover valid targets.
 
-The built-in mappings point every Claude spelling at the newest generally
-available Opus — currently `claude-opus-5` — and every Haiku spelling at
+The built-in mappings point every Opus spelling at the newest generally
+available Opus — currently `claude-opus-5` — every Sonnet spelling at the
+newest Sonnet (`claude-sonnet-5`), and every Haiku spelling at
 `claude-haiku-4.5`. Anthropic writes the same version two ways (`4.8` and
 `4-8`), so both forms are listed.
 
@@ -105,26 +98,6 @@ Controls the upstream base URL only:
 
 Set this to match the Copilot seat your token actually has.
 
-### GitHub Models
-
-[GitHub Models](https://models.github.ai) is GitHub's OpenAI-compatible model
-**inference** service, separate from Copilot. When `github_models.enabled` is
-true (the default), any request whose *translated* model id uses the
-`publisher/model` form (contains a `/`, e.g. `openai/gpt-4o`) is routed there
-instead of Copilot. These ids never collide with Copilot ids, so mappings and
-existing behavior are unaffected. Routing applies to `/v1/chat/completions`,
-`/v1/messages` (translated), and the Gemini endpoints.
-
-GitHub Models authenticates with the **raw GitHub token** (not the Copilot
-token) via `Authorization: Bearer`. That token must carry the **`models: read`**
-permission (a fine-grained PAT). This is **not** covered by the Device Flow login
-(`models` is not a valid classic OAuth scope), so supply a dedicated token via
-`github_models.token` (or `GHC_PROXY_GITHUB_MODELS_TOKEN`); otherwise GitHub
-Models requests fall back to the Device Flow token, which lacks this permission.
-The [setup wizard](getting-started.md#the-setup-wizard) can capture and validate
-this token for you. Set `github_models.org` to attribute inference to an
-organization. The catalog is merged into `GET /v1/models`.
-
 ### Schema upgrades
 
 `config_version` records the schema an existing `config.yaml` was written
@@ -147,6 +120,8 @@ Schema versions so far:
 | 2 | Opus 4.8 aliases |
 | 3 | `upstream_read_timeout_seconds`; `auto_upgrade` defaulting to true |
 | 4 | Opus 5 and Sonnet 5 aliases |
+| 5 | Gemini CLI model mappings (`gemini-*` prefixes) |
+| 6 | Removed the retired GitHub Models settings from persisted configuration |
 
 `--update-config` remains for the other, non-schema write-backs (for example
 restoring the built-in `model_mappings` when the file has none).
@@ -208,9 +183,6 @@ Every config field has a `GHC_PROXY_*` override:
 | `GHC_PROXY_RATE_LIMIT_WAIT` | Wait instead of rejecting when limited (`true`/`1`) |
 | `GHC_PROXY_MANUAL_APPROVE` | Require manual approval per request (`true`/`1`) |
 | `GHC_PROXY_API_KEY` | Require this key on LLM endpoints (empty = disabled) |
-| `GHC_PROXY_GITHUB_MODELS_ENABLED` | Route `publisher/model` ids to GitHub Models (`true`/`1`) |
-| `GHC_PROXY_GITHUB_MODELS_ORG` | Attribute GitHub Models inference to an organization |
-| `GHC_PROXY_GITHUB_MODELS_TOKEN` | Dedicated token for GitHub Models (`models: read` permission) |
 
 Token-related variables (`COPILOT_GITHUB_TOKEN`, `GH_TOKEN`, `GITHUB_TOKEN`) are
 covered in [Getting Started](getting-started.md#authentication).

docs/getting-started.md

@@ -71,11 +71,7 @@ The GitHub token is exchanged for a short-lived **Copilot token** via
 `https://api.github.com/copilot_internal/v2/token`, which the proxy refreshes
 automatically before it expires.
 
-The Device Flow requests the `read:user copilot` scopes. The
-[GitHub Models](configuration.md#github-models) inference API is not covered by
-the Device Flow token (`models` is not a valid classic OAuth scope); to use it,
-supply a dedicated token with the `models: read` permission (a fine-grained PAT)
-via `github_models.token`.
+The Device Flow requests the `read:user copilot` scopes.
 
 To authenticate without starting the server (useful for CI/headless setups):
 
@@ -98,12 +94,7 @@ It walks through:
 3. **Model mappings** — fetches the live model catalog and lets you map the
    `opus` / `sonnet` / `haiku` aliases to specific models, or keep the
    recommended defaults.
-4. **GitHub Models** — optionally enable routing of `publisher/model` ids to the
-   [GitHub Models](configuration.md#github-models) inference API. The wizard
-   checks whether your GitHub token already grants access and, if not, guides you
-   to create a fine-grained PAT with the `models: read` permission, validates the
-   token against the catalog, and saves it to `github_models.token`.
-5. **Client setup** — optionally configure Claude Code
+4. **Client setup** — optionally configure Claude Code
    (`~/.claude/settings.json`), Codex (`~/.codex/config.toml`), and the Gemini
    CLI (`~/.gemini/.env`) to route through the proxy. Existing settings are
    preserved and any user-set API key is left untouched.

public/dashboard.html

@@ -86,9 +86,16 @@ <h2>Prompt cache</h2>
          it cannot dominate the page you open to check spend. -->
     <details class="panel fold">
       <summary id="modelsSummary">Supported models</summary>
-      <div class="panel-body flush scroll-y">
+      <div class="panel-body flush scroll-y scroll-x">
         <table>
-          <thead><tr><th>ID</th><th>Display name</th><th>Owned by</th></tr></thead>
+          <thead>
+            <tr>
+              <th>ID</th><th>Display name</th><th>Owned by</th>
+              <th class="num">Context window</th>
+              <th class="num">Max output</th>
+              <th class="num">1M context</th>
+            </tr>
+          </thead>
           <tbody id="modelsBody"></tbody>
         </table>
       </div>
@@ -493,9 +500,23 @@ <h2>Prompt cache</h2>
       const data = await (await fetch('/api/models')).json();
       const models = (data && data.data) || [];
       document.getElementById('modelsSummary').textContent = `Supported models (${models.length})`;
-      document.getElementById('modelsBody').innerHTML = models.map(m =>
-        `<tr><td class="mono">${esc(m.id)}</td><td>${esc(m.display_name)}</td><td class="muted">${esc(m.owned_by)}</td></tr>`
-      ).join('');
+      document.getElementById('modelsBody').innerHTML = models.map(m => {
+        const context = m.context_window == null ? '—' : compact(m.context_window);
+        const output = m.max_output_tokens == null ? '—' : compact(m.max_output_tokens);
+        const extended = m.supports_1m_context == null
+          ? '<span class="muted">—</span>'
+          : m.supports_1m_context
+            ? '<span class="status-ok">yes</span>'
+            : '<span class="muted">no</span>';
+        return `<tr>
+          <td class="mono">${esc(m.id)}</td>
+          <td>${esc(m.display_name)}</td>
+          <td class="muted">${esc(m.owned_by)}</td>
+          <td class="num">${context}</td>
+          <td class="num">${output}</td>
+          <td class="num">${extended}</td>
+        </tr>`;
+      }).join('');
     }
 
     async function refresh() {