feat(vscode-lm): re-apply image support with types version fix

Re-applies the image support feature from PR #11065 that was reverted,
with a fix for the @types/vscode version issue.

The fix keeps @types/vscode at ^1.84.0 in package.json (matching
engines.vscode) while the pnpm-lock.yaml resolves to version 1.108.1
which provides the LanguageModelDataPart API.

This approach:
- Maintains compatibility with VS Code 1.84.0+
- Provides TypeScript types for the new LanguageModelDataPart API
- Resolves the vsix build failure from the original PR

Changes:
- Add convertImageToDataPart() for image-to-LanguageModelDataPart conversion
- Add checkModelSupportsImages() for model image capability detection
- Update useSelectedModel to use models actual supportsImages capability
- Add comprehensive tests for image support

Relates to #11064, #11065

Author: roomote

pnpm-lock.yaml

@@ -1,4 +1,5 @@
 import type { Mock } from "vitest"
+import { checkModelSupportsImages, IMAGE_CAPABLE_MODEL_PREFIXES } from "../vscode-lm"
 
 // Mocks must come first, before imports
 vi.mock("vscode", () => {
@@ -537,3 +538,92 @@ describe("VsCodeLmHandler", () => {
 		})
 	})
 })
+
+describe("checkModelSupportsImages", () => {
+	describe("OpenAI GPT models", () => {
+		it("should return true for all gpt-* models (GitHub Copilot)", () => {
+			// All GPT models in GitHub Copilot support images
+			expect(checkModelSupportsImages("gpt", "gpt-4o")).toBe(true)
+			expect(checkModelSupportsImages("gpt", "gpt-4.1")).toBe(true)
+			expect(checkModelSupportsImages("gpt", "gpt-5")).toBe(true)
+			expect(checkModelSupportsImages("gpt", "gpt-5.1")).toBe(true)
+			expect(checkModelSupportsImages("gpt", "gpt-5.2")).toBe(true)
+			expect(checkModelSupportsImages("gpt-mini", "gpt-5-mini")).toBe(true)
+			expect(checkModelSupportsImages("gpt-codex", "gpt-5.1-codex")).toBe(true)
+			expect(checkModelSupportsImages("gpt-codex", "gpt-5.2-codex")).toBe(true)
+			expect(checkModelSupportsImages("gpt-codex", "gpt-5.1-codex-max")).toBe(true)
+			expect(checkModelSupportsImages("gpt-codex", "gpt-5.1-codex-mini")).toBe(true)
+		})
+
+		it("should return true for o1 and o3 reasoning models", () => {
+			expect(checkModelSupportsImages("o1", "o1-preview")).toBe(true)
+			expect(checkModelSupportsImages("o1", "o1-mini")).toBe(true)
+			expect(checkModelSupportsImages("o3", "o3")).toBe(true)
+		})
+	})
+
+	describe("Anthropic Claude models", () => {
+		it("should return true for all claude-* models (GitHub Copilot)", () => {
+			// All Claude models in GitHub Copilot support images
+			expect(checkModelSupportsImages("claude-haiku", "claude-haiku-4.5")).toBe(true)
+			expect(checkModelSupportsImages("claude-opus", "claude-opus-4.5")).toBe(true)
+			expect(checkModelSupportsImages("claude-sonnet", "claude-sonnet-4")).toBe(true)
+			expect(checkModelSupportsImages("claude-sonnet", "claude-sonnet-4.5")).toBe(true)
+		})
+	})
+
+	describe("Google Gemini models", () => {
+		it("should return true for all gemini-* models (GitHub Copilot)", () => {
+			// All Gemini models in GitHub Copilot support images
+			expect(checkModelSupportsImages("gemini-pro", "gemini-2.5-pro")).toBe(true)
+			expect(checkModelSupportsImages("gemini-flash", "gemini-3-flash-preview")).toBe(true)
+			expect(checkModelSupportsImages("gemini-pro", "gemini-3-pro-preview")).toBe(true)
+		})
+	})
+
+	describe("non-vision models", () => {
+		it("should return false for grok models (text-only in GitHub Copilot)", () => {
+			// Grok is the only model family in GitHub Copilot that doesn't support images
+			expect(checkModelSupportsImages("grok", "grok-code-fast-1")).toBe(false)
+		})
+
+		it("should return false for models with non-matching prefixes", () => {
+			// Models that don't start with gpt, claude, gemini, o1, or o3
+			expect(checkModelSupportsImages("mistral", "mistral-large")).toBe(false)
+			expect(checkModelSupportsImages("llama", "llama-3-70b")).toBe(false)
+			expect(checkModelSupportsImages("unknown", "some-random-model")).toBe(false)
+		})
+	})
+
+	describe("case insensitivity", () => {
+		it("should match regardless of case", () => {
+			expect(checkModelSupportsImages("GPT", "GPT-4O")).toBe(true)
+			expect(checkModelSupportsImages("CLAUDE", "CLAUDE-SONNET-4")).toBe(true)
+			expect(checkModelSupportsImages("GEMINI", "GEMINI-2.5-PRO")).toBe(true)
+		})
+	})
+
+	describe("prefix matching", () => {
+		it("should only match IDs that start with known prefixes", () => {
+			// ID must START with the prefix, not just contain it
+			expect(checkModelSupportsImages("custom", "gpt-4o")).toBe(true) // ID starts with gpt
+			expect(checkModelSupportsImages("custom", "my-gpt-model")).toBe(false) // gpt not at start
+			expect(checkModelSupportsImages("custom", "not-claude-model")).toBe(false) // claude not at start
+		})
+	})
+})
+
+describe("IMAGE_CAPABLE_MODEL_PREFIXES", () => {
+	it("should export the model prefixes array", () => {
+		expect(Array.isArray(IMAGE_CAPABLE_MODEL_PREFIXES)).toBe(true)
+		expect(IMAGE_CAPABLE_MODEL_PREFIXES.length).toBeGreaterThan(0)
+	})
+
+	it("should include key model prefixes", () => {
+		expect(IMAGE_CAPABLE_MODEL_PREFIXES).toContain("gpt")
+		expect(IMAGE_CAPABLE_MODEL_PREFIXES).toContain("claude")
+		expect(IMAGE_CAPABLE_MODEL_PREFIXES).toContain("gemini")
+		expect(IMAGE_CAPABLE_MODEL_PREFIXES).toContain("o1")
+		expect(IMAGE_CAPABLE_MODEL_PREFIXES).toContain("o3")
+	})
+})

src/api/providers/__tests__/vscode-lm.spec.ts

@@ -529,14 +529,18 @@ export class VsCodeLmHandler extends BaseProvider implements SingleCompletionHan
 
 			const modelId = this.client.id || modelParts.join(SELECTOR_SEPARATOR)
 
+			// Check if the model supports images based on known model families
+			// VS Code Language Model API 1.106+ supports image inputs via LanguageModelDataPart
+			const supportsImages = checkModelSupportsImages(this.client.family, this.client.id)
+
 			// Build model info with conservative defaults for missing values
 			const modelInfo: ModelInfo = {
 				maxTokens: -1, // Unlimited tokens by default
 				contextWindow:
 					typeof this.client.maxInputTokens === "number"
 						? Math.max(0, this.client.maxInputTokens)
 						: openAiModelInfoSaneDefaults.contextWindow,
-				supportsImages: false, // VSCode Language Model API currently doesn't support image inputs
+				supportsImages,
 				supportsPromptCache: true,
 				inputPrice: 0,
 				outputPrice: 0,
@@ -586,8 +590,43 @@ export class VsCodeLmHandler extends BaseProvider implements SingleCompletionHan
 	}
 }
 
-// Static blacklist of VS Code Language Model IDs that should be excluded from the model list e.g. because they will never work
-const VSCODE_LM_STATIC_BLACKLIST: string[] = ["claude-3.7-sonnet", "claude-3.7-sonnet-thought"]
+/**
+ * Model ID prefixes that support image inputs via VS Code Language Model API.
+ * These models support the LanguageModelDataPart.image() API introduced in VS Code 1.106+.
+ *
+ * All GitHub Copilot models with these prefixes support images.
+ * Only grok-* models don't support images (text only).
+ *
+ * Source: https://models.dev/api.json (github-copilot provider models)
+ */
+export const IMAGE_CAPABLE_MODEL_PREFIXES = [
+	"gpt", // All GPT models (gpt-4o, gpt-4.1, gpt-5, gpt-5.1, gpt-5.2, gpt-5-mini, gpt-5.1-codex, etc.)
+	"claude", // All Claude models (claude-haiku-4.5, claude-opus-4.5, claude-sonnet-4, claude-sonnet-4.5)
+	"gemini", // All Gemini models (gemini-2.5-pro, gemini-3-flash-preview, gemini-3-pro-preview)
+	"o1", // OpenAI o1 reasoning models
+	"o3", // OpenAI o3 reasoning models
+]
+
+/**
+ * Checks if a model supports image inputs based on its model ID.
+ * Uses prefix matching against known image-capable model families.
+ *
+ * @param _family The model family (unused, kept for API compatibility)
+ * @param id The model ID
+ * @returns true if the model supports image inputs
+ */
+export function checkModelSupportsImages(_family: string, id: string): boolean {
+	const idLower = id.toLowerCase()
+	return IMAGE_CAPABLE_MODEL_PREFIXES.some((prefix) => idLower.startsWith(prefix))
+}
+
+// Static blacklist of VS Code Language Model IDs that should be excluded from the model list
+// e.g. because they don't support native tool calling or will never work
+const VSCODE_LM_STATIC_BLACKLIST: string[] = [
+	"claude-3.7-sonnet",
+	"claude-3.7-sonnet-thought",
+	"claude-opus-41", // Does not support native tool calling
+]
 
 export async function getVsCodeLmModels() {
 	try {