🤖 feat: add support for Claude Opus 4.8

.github/workflows/auto-cleanup-fixup.yml

@@ -118,6 +118,6 @@ jobs:
           FAILED_RUN_ID: ${{ github.event.workflow_run.id }}
         run: |
           bunx mux@next run \
-            --model anthropic:claude-opus-4-7 \
+            --model anthropic:claude-opus-4-8 \
             --thinking xhigh \
             < .github/prompts/auto-cleanup-fixup.md

.github/workflows/auto-cleanup.yml

@@ -71,6 +71,6 @@ jobs:
           GH_TOKEN: ${{ steps.app-token.outputs.token }}
         run: |
           bunx mux@next run \
-            --model anthropic:claude-opus-4-7 \
+            --model anthropic:claude-opus-4-8 \
             --thinking xhigh \
             < .github/prompts/auto-cleanup.md

.github/workflows/nightly-terminal-bench.yml

@@ -10,7 +10,7 @@ on:
   workflow_dispatch:
     inputs:
       models:
-        description: 'Models to test (comma-separated, or "all" for opus-4-7 + gpt-5.5 + google/gemini-3-pro-preview + google/gemini-3-flash-preview + google/gemini-3.5-flash)'
+        description: 'Models to test (comma-separated, or "all" for opus-4-8 + gpt-5.5 + google/gemini-3-pro-preview + google/gemini-3-flash-preview + google/gemini-3.5-flash)'
         required: false
         default: "all"
         type: string
@@ -129,7 +129,7 @@ jobs:
           INPUT_MODELS: ${{ inputs.models }}
         run: |
           if [ "$INPUT_MODELS" = "all" ] || [ -z "$INPUT_MODELS" ]; then
-            echo 'models=["anthropic/claude-opus-4-7","openai/gpt-5.5","google/gemini-3-pro-preview","google/gemini-3-flash-preview","google/gemini-3.5-flash"]' >> "$GITHUB_OUTPUT"
+            echo 'models=["anthropic/claude-opus-4-8","openai/gpt-5.5","google/gemini-3-pro-preview","google/gemini-3-flash-preview","google/gemini-3.5-flash"]' >> "$GITHUB_OUTPUT"
           else
             # Convert comma-separated to JSON array
             models_json=$(echo "$INPUT_MODELS" | jq -R -s -c 'split(",") | map(gsub("^\\s+|\\s+$"; ""))')

.github/workflows/terminal-bench.yml

@@ -4,7 +4,7 @@ on:
   workflow_call:
     inputs:
       model_name:
-        description: "Model to use (e.g., anthropic/claude-opus-4-7, openai/gpt-5.5)"
+        description: "Model to use (e.g., anthropic/claude-opus-4-8, openai/gpt-5.5)"
         required: false
         type: string
       dataset:
@@ -98,7 +98,7 @@ on:
         required: false
         type: string
       model_name:
-        description: "Model to use (e.g., anthropic/claude-opus-4-7, openai/gpt-5.5)"
+        description: "Model to use (e.g., anthropic/claude-opus-4-8, openai/gpt-5.5)"
         required: false
         type: string
       mux_run_args:

docs/config/models.mdx

@@ -13,7 +13,7 @@ Mux ships with curated models kept up to date with the frontier. Use any custom
 
 | Model                  | ID                            | Aliases                                                      | Default |
 | ---------------------- | ----------------------------- | ------------------------------------------------------------ | ------- |
-| Opus 4.7               | anthropic:claude-opus-4-7     | `opus`                                                       | ✓       |
+| Opus 4.8               | anthropic:claude-opus-4-8     | `opus`                                                       | ✓       |
 | Sonnet 4.6             | anthropic:claude-sonnet-4-6   | `sonnet`                                                     |         |
 | Haiku 4.5              | anthropic:claude-haiku-4-5    | `haiku`                                                      |         |
 | GPT-5.5                | openai:gpt-5.5                | `gpt`, `gpt-5.5`                                             |         |

docs/guides/github-actions.mdx

@@ -116,7 +116,7 @@ jobs:
           GH_TOKEN: ${{ steps.app-token.outputs.token }}
         run: |
           bunx mux@next run \
-            --model anthropic:claude-opus-4-7 \
+            --model anthropic:claude-opus-4-8 \
             --thinking xhigh \
             < .github/prompts/auto-cleanup.md
 ```

src/browser/features/ChatInput/useCreationWorkspace.test.tsx

@@ -801,7 +801,7 @@ describe("useCreationWorkspace", () => {
     });
     expect(onWorkspaceCreated.mock.calls[0][1]).toEqual({
       autoNavigate: true,
-      pendingStreamModel: "anthropic:claude-opus-4-7",
+      pendingStreamModel: "anthropic:claude-opus-4-8",
       markPendingInitialSend: false,
     });
   });
@@ -1196,7 +1196,7 @@ describe("useCreationWorkspace", () => {
     expect(onWorkspaceCreated.mock.calls.length).toBe(1);
     expect(onWorkspaceCreated.mock.calls[0][1]).toEqual({
       autoNavigate: true,
-      pendingStreamModel: "anthropic:claude-opus-4-7",
+      pendingStreamModel: "anthropic:claude-opus-4-8",
       markPendingInitialSend: true,
     });
   });

src/browser/hooks/useModelsFromSettings.test.ts

@@ -586,7 +586,7 @@ describe("useModelsFromSettings provider availability gating", () => {
         apiKeySet: true,
         isEnabled: true,
         isConfigured: true,
-        models: ["claude-opus-4.7"],
+        models: ["claude-opus-4.8"],
       },
     };
     routePriority = ["github-copilot", "direct"];

src/cli/run.test.ts

@@ -167,7 +167,7 @@ describe("mux CLI", () => {
     test("shows default model as opus", async () => {
       const result = await runCli(["run", "--help"]);
       expect(result.exitCode).toBe(0);
-      expect(result.stdout).toContain("anthropic:claude-opus-4-7");
+      expect(result.stdout).toContain("anthropic:claude-opus-4-8");
     });
 
     test("--service-tier has no default auto", async () => {

src/cli/run.ts

@@ -91,7 +91,7 @@ import type { GoalRecordV1 } from "@/common/types/goal";
 
 // Display labels for CLI help (OFF, LOW, MED, HIGH, MAX).
 // Deduplicate because xhigh and max both display as "MAX" for default/Anthropic
-// models; the distinction only matters at the provider boundary (OpenAI / Opus 4.7).
+// models; the distinction only matters at the provider boundary (OpenAI / Opus 4.7+).
 const THINKING_LABELS_LIST = [...new Set(Object.values(THINKING_DISPLAY_LABELS))].join(", ");
 
 type CLIMode = "plan" | "exec";

src/common/constants/knownModels.ts

@@ -29,10 +29,11 @@ interface KnownModel extends KnownModelDefinition {
 const MODEL_DEFINITIONS = {
   OPUS: {
     provider: "anthropic",
-    providerModelId: "claude-opus-4-7",
+    providerModelId: "claude-opus-4-8",
     aliases: ["opus"],
     warm: true,
-    // Opus 4.7 tokenizer not yet available upstream; reuse 4.6 for approximate counting
+    // Opus 4.8 tokenizer not yet available upstream; reuse 4.5 for approximate counting
+    // (Opus 4.6/4.7 also reused 4.5 — tokenization is unchanged across the 4.x line).
     tokenizerOverride: "anthropic/claude-opus-4.5",
   },
   SONNET: {

src/common/utils/ai/models.ts

@@ -120,6 +120,7 @@ export type Anthropic1MContextMode = "none" | "beta" | "native";
 
 const OPTIONAL_VERSION_SUFFIX = String.raw`(?:-(?:\d{8}|\d{4}-\d{2}-\d{2}))?`;
 const ANTHROPIC_NATIVE_1M_PATTERNS = [
+  new RegExp(`^claude-opus-4-8${OPTIONAL_VERSION_SUFFIX}$`, "i"),
   new RegExp(`^claude-opus-4-7${OPTIONAL_VERSION_SUFFIX}$`, "i"),
   new RegExp(`^claude-opus-4-6${OPTIONAL_VERSION_SUFFIX}$`, "i"),
   new RegExp(`^claude-sonnet-4-6${OPTIONAL_VERSION_SUFFIX}$`, "i"),

src/common/utils/tokens/models-extra.ts

@@ -55,6 +55,27 @@ export const modelsExtra: Record<string, ModelData> = {
     supports_pdf_input: true,
   },
 
+  // Claude Opus 4.8 - Released May 28, 2026
+  // Same pricing/shape as Opus 4.7: $5/M input, $25/M output, native 1M context,
+  // 128K max output, native xhigh effort level. Defaults to high effort on all
+  // surfaces; "fast mode" (separate provider tier) is billed at 2× standard rates
+  // ($10/$50) and is not exposed as its own first-class entry yet.
+  "claude-opus-4-8": {
+    max_input_tokens: 1000000,
+    max_output_tokens: 128000,
+    input_cost_per_token: 0.000005, // $5 per million input tokens
+    output_cost_per_token: 0.000025, // $25 per million output tokens
+    cache_creation_input_token_cost: 0.00000625, // $6.25 per million tokens
+    cache_read_input_token_cost: 0.0000005, // $0.50 per million tokens
+    litellm_provider: "anthropic",
+    mode: "chat",
+    supports_function_calling: true,
+    supports_vision: true,
+    supports_pdf_input: true,
+    supports_reasoning: true,
+    supports_response_schema: true,
+  },
+
   // Claude Opus 4.7 - Released April 2026
   // Native 1M context at standard pricing: $5/M input, $25/M output.
   // 128K max output tokens. Supports native xhigh effort level.

src/node/services/agentSkills/builtInSkillContent.generated.ts

@@ -2376,7 +2376,7 @@ export const BUILTIN_SKILL_FILES: Record<string, Record<string, string>> = {
       "",
       "| Model                  | ID                            | Aliases                                                      | Default |",
       "| ---------------------- | ----------------------------- | ------------------------------------------------------------ | ------- |",
-      "| Opus 4.7               | anthropic:claude-opus-4-7     | `opus`                                                       | ✓       |",
+      "| Opus 4.8               | anthropic:claude-opus-4-8     | `opus`                                                       | ✓       |",
       "| Sonnet 4.6             | anthropic:claude-sonnet-4-6   | `sonnet`                                                     |         |",
       "| Haiku 4.5              | anthropic:claude-haiku-4-5    | `haiku`                                                      |         |",
       "| GPT-5.5                | openai:gpt-5.5                | `gpt`, `gpt-5.5`                                             |         |",
@@ -3834,7 +3834,7 @@ export const BUILTIN_SKILL_FILES: Record<string, Record<string, string>> = {
       "          GH_TOKEN: ${{ steps.app-token.outputs.token }}",
       "        run: |",
       "          bunx mux@next run \\",
-      "            --model anthropic:claude-opus-4-7 \\",
+      "            --model anthropic:claude-opus-4-8 \\",
       "            --thinking xhigh \\",
       "            < .github/prompts/auto-cleanup.md",
       "```",

src/node/services/providerService.test.ts

@@ -1169,7 +1169,7 @@ describe("ProviderService.setConfig", () => {
       const providersConfig = config.loadProvidersConfig();
       expect(providersConfig?.["mux-gateway"]?.models).toEqual([
         "anthropic/claude-sonnet-4-6",
-        "anthropic/claude-opus-4-7",
+        "anthropic/claude-opus-4-8",
         "openai/gpt-5.5",
       ]);
       expect(providersConfig?.["mux-gateway"]?.models).not.toContain("openai/gpt-5.2-codex");

src/node/services/providerService.ts

@@ -1075,7 +1075,7 @@ export class ProviderService {
         if (existingModels.length === 0) {
           providerConfig.models = [
             "anthropic/claude-sonnet-4-6",
-            "anthropic/claude-opus-4-7",
+            "anthropic/claude-opus-4-8",
             "openai/gpt-5.5",
           ];
         }

tests/e2e/scenarios/slashCommands.spec.ts

@@ -76,8 +76,8 @@ test.describe("slash command flows", () => {
     await ui.projects.openFirstWorkspace();
 
     const modeToggles = page.locator('[data-component="ChatModeToggles"]');
-    // Default model is Opus 4.7 - displayed as formatted name
-    await expect(modeToggles.getByText("Opus 4.7", { exact: true })).toBeVisible();
+    // Default model is Opus 4.8 - displayed as formatted name
+    await expect(modeToggles.getByText("Opus 4.8", { exact: true })).toBeVisible();
 
     await ui.chat.sendMessage("/model sonnet");
     await ui.chat.expectStatusMessageContains("Model changed to anthropic:claude-sonnet-4-6");