Feat/multi model provider support (#692)

* test: add tests for provider model env updates and multi-model profiles Add comprehensive tests covering: - OPENAI_MODEL/ANTHROPIC_MODEL env updates on provider activation - Cross-provider type switches (openai ↔ anthropic) clearing stale env - Multi-model profile activation using only the first model for env vars - Model options cache population from comma-separated model lists - getProfileModelOptions generating correct ModelOption arrays * feat: multi-model provider support and model auto-switch Support comma-separated model names in provider profiles (e.g. "glm-4.7, glm-4.7-flash"). The first model is used as default on activation; all models appear in the /model picker for easy switching. When switching active providers, the session model now automatically updates to the new provider's first model. The multi-model list is preserved across switches and /model selections. Changes: - Add parseModelList, getPrimaryModel, hasMultipleModels utilities with full test coverage (19 tests) - Use getPrimaryModel when applying profiles to process.env so only the primary model is set in OPENAI_MODEL/ANTHROPIC_MODEL - Update ProviderManager UI to hint at multi-model syntax and show model count in provider list summaries - Populate model options cache from multi-model profiles on activation so all models appear in /model picker regardless of base URL type - Guard persistActiveProviderProfileModel against overwriting comma-separated lists: models already in the profile are session selections, not profile edits - Set AppState.mainLoopModel to the actual model string on provider switch so Anthropic profiles use the configured model instead of falling back to the built-in default * fix: only show profile models when provider profile env is applied Guard the profile model picker options behind a PROFILE_ENV_APPLIED check. getActiveProviderProfile() has a ?? profiles[0] fallback that returns the first profile even when no profile is explicitly active, causing users with inactive profiles to lose all standard model options (Opus, Haiku, etc.) from the /model picker. * fix: show all model names for profiles with 3 or fewer models Instead of a summary format for multi-model profiles, display all model names when there are 3 or fewer. Only use the "+ N more" format for profiles with 4+ models. * fix: preserve standard model options in picker alongside profile models The previous implementation used an early return that replaced all standard picker options (Opus, Haiku, Sonnet for Anthropic; Codex/GPT models for OpenAI) with only the profile's custom models. Changes: - Collect profile models into a shared array instead of early returning - Append profile models to firstParty path (Opus + Haiku + Sonnet + custom) - Append profile models to PAYG 3P path (Codex + Sonnet + Opus + Haiku + custom) - Guard collection behind PROFILE_ENV_APPLIED to avoid ?? profiles[0] fallback Fixes review feedback: standard models are no longer hidden when a provider profile with custom models is active. Users see both the standard options and their profile's models. --------- Co-authored-by: Ali Alakbarli <ali.alakbarli@users.noreply.github.com>
2026-04-16 01:01:55 +04:00
parent 51191d6132
commit b66633ea4d
6 changed files with 505 additions and 9 deletions
--- a/src/components/ProviderManager.tsx
+++ b/src/components/ProviderManager.tsx
@@ -9,6 +9,7 @@ import {
  readCodexCredentialsAsync,
 } from '../utils/codexCredentials.js'
 import { isBareMode, isEnvTruthy } from '../utils/envUtils.js'
+import { getPrimaryModel, hasMultipleModels, parseModelList } from '../utils/providerModels.js'
 import {
  applySavedProfileToCurrentSession,
  buildCodexOAuthProfileEnv,
@@ -50,6 +51,7 @@ import {
 import { Pane } from './design-system/Pane.js'
 import TextInput from './TextInput.js'
 import { useCodexOAuthFlow } from './useCodexOAuthFlow.js'
+import { useSetAppState } from '../state/AppState.js'

 export type ProviderManagerResult = {
  action: 'saved' | 'cancelled'
@@ -108,8 +110,8 @@ const FORM_STEPS: Array<{
  {
    key: 'model',
    label: 'Default model',
-    placeholder: 'e.g. llama3.1:8b',
-    helpText: 'Model name to use when this provider is active.',
+    placeholder: 'e.g. llama3.1:8b or glm-4.7, glm-4.7-flash',
+    helpText: 'Model name(s) to use. Separate multiple with commas; first is default.',
  },
  {
    key: 'apiKey',
@@ -153,7 +155,12 @@ function profileSummary(profile: ProviderProfile, isActive: boolean): string {
  const keyInfo = profile.apiKey ? 'key set' : 'no key'
  const providerKind =
    profile.provider === 'anthropic' ? 'anthropic' : 'openai-compatible'
-  return `${providerKind} · ${profile.baseUrl} · ${profile.model} · ${keyInfo}${activeSuffix}`
+  const models = parseModelList(profile.model)
+  const modelDisplay =
+    models.length <= 3
+      ? models.join(', ')
+      : `${models[0]}, ${models[1]} + ${models.length - 2} more`
+  return `${providerKind} · ${profile.baseUrl} · ${modelDisplay} · ${keyInfo}${activeSuffix}`
 }

 function getGithubCredentialSourceFromEnv(
@@ -320,6 +327,7 @@ function CodexOAuthSetup({
 }

 export function ProviderManager({ mode, onDone }: Props): React.ReactNode {
+  const setAppState = useSetAppState()
  const initialGithubCredentialSource = getGithubCredentialSourceFromEnv()
  const initialIsGithubActive = isEnvTruthy(process.env.CLAUDE_CODE_USE_GITHUB)
  const initialHasGithubCredential = initialGithubCredentialSource !== 'none'
@@ -573,6 +581,10 @@ export function ProviderManager({ mode, onDone }: Props): React.ReactNode {
        }

        refreshProfiles()
+        setAppState(prev => ({
+          ...prev,
+          mainLoopModel: GITHUB_PROVIDER_DEFAULT_MODEL,
+        }))
        setStatusMessage(`Active provider: ${GITHUB_PROVIDER_LABEL}`)
        setScreen('menu')
        return
@@ -585,6 +597,16 @@ export function ProviderManager({ mode, onDone }: Props): React.ReactNode {
        return
      }

+      // Update the session model to the new provider's first model.
+      // persistActiveProviderProfileModel (called by onChangeAppState) will
+      // not overwrite the multi-model list because it checks if the model
+      // is already in the profile's comma-separated model list.
+      const newModel = getPrimaryModel(active.model)
+      setAppState(prev => ({
+        ...prev,
+        mainLoopModel: newModel,
+      }))
+
      providerLabel = active.name
      const settingsOverrideError =
        clearStartupProviderOverrideFromUserSettings()
--- a/src/utils/model/modelOptions.ts
+++ b/src/utils/model/modelOptions.ts
@@ -33,7 +33,11 @@ import {
 } from './model.js'
 import { has1mContext } from '../context.js'
 import { getGlobalConfig } from '../config.js'
-import { getActiveOpenAIModelOptionsCache } from '../providerProfiles.js'
+import {
+  getActiveOpenAIModelOptionsCache,
+  getActiveProviderProfile,
+  getProfileModelOptions,
+} from '../providerProfiles.js'
 import { getCachedOllamaModelOptions, isOllamaProvider } from './ollamaModels.js'
 import { getCachedNvidiaNimModelOptions, isNvidiaNimProvider } from './nvidiaNimModels.js'
 import { getCachedMiniMaxModelOptions, isMiniMaxProvider } from './minimaxModels.js'
@@ -476,6 +480,20 @@ function getModelOptionsBase(fastMode = false): ModelOption[] {
    ]
  }

+  // When a provider profile's env is applied, collect its models so they
+  // can be appended to the standard picker options below.
+  // We check PROFILE_ENV_APPLIED to avoid the ?? profiles[0] fallback in
+  // getActiveProviderProfile which would affect users with inactive profiles.
+  const profileEnvApplied = process.env.CLAUDE_CODE_PROVIDER_PROFILE_ENV_APPLIED === '1'
+  const profileModelOptions: ModelOption[] = []
+  if (profileEnvApplied) {
+    const activeProfile = getActiveProviderProfile()
+    if (activeProfile) {
+      const models = getProfileModelOptions(activeProfile)
+      profileModelOptions.push(...models)
+    }
+  }
+
  // PAYG 1P API: Default (Sonnet) + Sonnet 1M + Opus 4.6 + Opus 1M + Haiku
  if (getAPIProvider() === 'firstParty') {
    const payg1POptions = [getDefaultOptionForUser(fastMode)]
@@ -491,6 +509,7 @@ function getModelOptionsBase(fastMode = false): ModelOption[] {
      }
    }
    payg1POptions.push(getHaiku45Option())
+    payg1POptions.push(...profileModelOptions)
    return payg1POptions
  }

@@ -530,6 +549,7 @@ function getModelOptionsBase(fastMode = false): ModelOption[] {
  } else {
    payg3pOptions.push(getHaikuOption())
  }
+  payg3pOptions.push(...profileModelOptions)
  return payg3pOptions
 }

--- a/src/utils/providerModels.test.ts
+++ b/src/utils/providerModels.test.ts
@@ -0,0 +1,108 @@
+import { describe, expect, test } from 'bun:test'
+
+import {
+  getPrimaryModel,
+  hasMultipleModels,
+  parseModelList,
+} from './providerModels.ts'
+
+// ── parseModelList ────────────────────────────────────────────────────────────
+
+describe('parseModelList', () => {
+  test('splits comma-separated models', () => {
+    expect(parseModelList('glm-4.7, glm-4.7-flash')).toEqual([
+      'glm-4.7',
+      'glm-4.7-flash',
+    ])
+  })
+
+  test('returns single model in an array', () => {
+    expect(parseModelList('llama3.1:8b')).toEqual(['llama3.1:8b'])
+  })
+
+  test('trims whitespace around each model', () => {
+    expect(parseModelList('  gpt-4o ,  gpt-4o-mini  , o3-mini ')).toEqual([
+      'gpt-4o',
+      'gpt-4o-mini',
+      'o3-mini',
+    ])
+  })
+
+  test('filters out empty entries from trailing commas', () => {
+    expect(parseModelList('gpt-4o,,gpt-4o-mini,')).toEqual([
+      'gpt-4o',
+      'gpt-4o-mini',
+    ])
+  })
+
+  test('returns empty array for empty string', () => {
+    expect(parseModelList('')).toEqual([])
+  })
+
+  test('returns empty array for whitespace-only string', () => {
+    expect(parseModelList('   ')).toEqual([])
+  })
+
+  test('returns empty array for comma-only string', () => {
+    expect(parseModelList(',,,')).toEqual([])
+  })
+
+  test('handles models with colons', () => {
+    expect(parseModelList('qwen2.5-coder:7b, llama3.1:8b')).toEqual([
+      'qwen2.5-coder:7b',
+      'llama3.1:8b',
+    ])
+  })
+})
+
+// ── getPrimaryModel ───────────────────────────────────────────────────────────
+
+describe('getPrimaryModel', () => {
+  test('returns first model from comma-separated list', () => {
+    expect(getPrimaryModel('glm-4.7, glm-4.7-flash')).toBe('glm-4.7')
+  })
+
+  test('returns the only model when single model is provided', () => {
+    expect(getPrimaryModel('llama3.1:8b')).toBe('llama3.1:8b')
+  })
+
+  test('returns the original string when input is empty', () => {
+    expect(getPrimaryModel('')).toBe('')
+  })
+
+  test('returns first model after trimming', () => {
+    expect(getPrimaryModel('  gpt-4o , gpt-4o-mini')).toBe('gpt-4o')
+  })
+
+  test('returns first model when others are empty from trailing commas', () => {
+    expect(getPrimaryModel('claude-sonnet-4-6,,')).toBe('claude-sonnet-4-6')
+  })
+})
+
+// ── hasMultipleModels ─────────────────────────────────────────────────────────
+
+describe('hasMultipleModels', () => {
+  test('returns true when multiple models are present', () => {
+    expect(hasMultipleModels('glm-4.7, glm-4.7-flash')).toBe(true)
+  })
+
+  test('returns false for a single model', () => {
+    expect(hasMultipleModels('llama3.1:8b')).toBe(false)
+  })
+
+  test('returns false for empty string', () => {
+    expect(hasMultipleModels('')).toBe(false)
+  })
+
+  test('returns false for whitespace-only string', () => {
+    expect(hasMultipleModels('   ')).toBe(false)
+  })
+
+  test('returns false when extra commas produce no extra models', () => {
+    expect(hasMultipleModels('gpt-4o,,')).toBe(false)
+  })
+
+  test('returns true for three models', () => {
+    expect(hasMultipleModels('a, b, c')).toBe(true)
+  })
+})
--- a/src/utils/providerModels.ts
+++ b/src/utils/providerModels.ts
@@ -0,0 +1,33 @@
+/**
+ * Utility functions for parsing comma-separated model names in provider profiles.
+ *
+ * Example: "glm-4.7, glm-4.7-flash" -> ["glm-4.7", "glm-4.7-flash"]
+ * Single model: "llama3.1:8b" -> ["llama3.1:8b"]
+ */
+
+/**
+ * Splits a comma-separated model field into an array of trimmed model names,
+ * filtering out any empty entries.
+ */
+export function parseModelList(modelField: string): string[] {
+  return modelField
+    .split(',')
+    .map((part) => part.trim())
+    .filter((part) => part.length > 0)
+}
+
+/**
+ * Returns the first (primary) model from a comma-separated model field.
+ * Falls back to the original string if parsing yields no results.
+ */
+export function getPrimaryModel(modelField: string): string {
+  const models = parseModelList(modelField)
+  return models.length > 0 ? models[0] : modelField
+}
+
+/**
+ * Returns true if the model field contains more than one model.
+ */
+export function hasMultipleModels(modelField: string): boolean {
+  return parseModelList(modelField).length > 1
+}
--- a/src/utils/providerProfiles.test.ts
+++ b/src/utils/providerProfiles.test.ts
@@ -139,6 +139,39 @@ describe('applyProviderProfileToProcessEnv', () => {
    expect(process.env.CLAUDE_CODE_USE_OPENAI).toBeUndefined()
    expect(getFreshAPIProvider()).toBe('firstParty')
  })
+
+  test('openai profile with multi-model string sets only first model in OPENAI_MODEL', async () => {
+    const { applyProviderProfileToProcessEnv } =
+      await importFreshProviderProfileModules()
+
+    applyProviderProfileToProcessEnv(
+      buildProfile({
+        provider: 'openai',
+        baseUrl: 'https://api.openai.com/v1',
+        model: 'glm-4.7, glm-4.7-flash, glm-4.7-plus',
+      }),
+    )
+
+    expect(process.env.OPENAI_MODEL).toBe('glm-4.7')
+    expect(String(process.env.CLAUDE_CODE_USE_OPENAI)).toBe('1')
+    expect(process.env.OPENAI_BASE_URL).toBe('https://api.openai.com/v1')
+  })
+
+  test('anthropic profile with multi-model string sets only first model in ANTHROPIC_MODEL', async () => {
+    const { applyProviderProfileToProcessEnv } =
+      await importFreshProviderProfileModules()
+
+    applyProviderProfileToProcessEnv(
+      buildProfile({
+        provider: 'anthropic',
+        baseUrl: 'https://api.anthropic.com',
+        model: 'claude-sonnet-4-6, claude-opus-4-6',
+      }),
+    )
+
+    expect(process.env.ANTHROPIC_MODEL).toBe('claude-sonnet-4-6')
+    expect(process.env.ANTHROPIC_BASE_URL).toBe('https://api.anthropic.com')
+  })
 })

 describe('applyActiveProviderProfileFromConfig', () => {
@@ -361,6 +394,169 @@ describe('getProviderPresetDefaults', () => {
  })
 })

+describe('setActiveProviderProfile', () => {
+  test('sets OPENAI_MODEL env var when switching to an openai-type provider', async () => {
+    const { setActiveProviderProfile } =
+      await importFreshProviderProfileModules()
+    const openaiProfile = buildProfile({
+      id: 'openai_prof',
+      name: 'OpenAI Provider',
+      provider: 'openai',
+      baseUrl: 'https://api.openai.com/v1',
+      model: 'gpt-4o',
+    })
+
+    saveMockGlobalConfig(current => ({
+      ...current,
+      providerProfiles: [openaiProfile],
+    }))
+
+    const result = setActiveProviderProfile('openai_prof')
+
+    expect(result?.id).toBe('openai_prof')
+    expect(String(process.env.CLAUDE_CODE_USE_OPENAI)).toBe('1')
+    expect(process.env.OPENAI_MODEL).toBe('gpt-4o')
+    expect(process.env.OPENAI_BASE_URL).toBe('https://api.openai.com/v1')
+    expect(process.env.CLAUDE_CODE_PROVIDER_PROFILE_ENV_APPLIED_ID).toBe(
+      'openai_prof',
+    )
+  })
+
+  test('sets ANTHROPIC_MODEL env var when switching to an anthropic-type provider', async () => {
+    const { setActiveProviderProfile } =
+      await importFreshProviderProfileModules()
+    const anthropicProfile = buildProfile({
+      id: 'anthro_prof',
+      name: 'Anthropic Provider',
+      provider: 'anthropic',
+      baseUrl: 'https://api.anthropic.com',
+      model: 'claude-sonnet-4-6',
+    })
+
+    saveMockGlobalConfig(current => ({
+      ...current,
+      providerProfiles: [anthropicProfile],
+    }))
+
+    const result = setActiveProviderProfile('anthro_prof')
+
+    expect(result?.id).toBe('anthro_prof')
+    expect(process.env.ANTHROPIC_MODEL).toBe('claude-sonnet-4-6')
+    expect(process.env.ANTHROPIC_BASE_URL).toBe('https://api.anthropic.com')
+    expect(process.env.CLAUDE_CODE_USE_OPENAI).toBeUndefined()
+    expect(process.env.OPENAI_MODEL).toBeUndefined()
+    expect(process.env.CLAUDE_CODE_PROVIDER_PROFILE_ENV_APPLIED_ID).toBe(
+      'anthro_prof',
+    )
+  })
+
+  test('clears openai model env and sets anthropic model env when switching from openai to anthropic provider', async () => {
+    const { setActiveProviderProfile } =
+      await importFreshProviderProfileModules()
+    const openaiProfile = buildProfile({
+      id: 'openai_prof',
+      name: 'OpenAI Provider',
+      provider: 'openai',
+      baseUrl: 'https://api.openai.com/v1',
+      model: 'gpt-4o',
+      apiKey: 'sk-openai-key',
+    })
+    const anthropicProfile = buildProfile({
+      id: 'anthro_prof',
+      name: 'Anthropic Provider',
+      provider: 'anthropic',
+      baseUrl: 'https://api.anthropic.com',
+      model: 'claude-sonnet-4-6',
+      apiKey: 'sk-ant-key',
+    })
+
+    saveMockGlobalConfig(current => ({
+      ...current,
+      providerProfiles: [openaiProfile, anthropicProfile],
+    }))
+
+    // First activate the openai profile
+    setActiveProviderProfile('openai_prof')
+    expect(process.env.OPENAI_MODEL).toBe('gpt-4o')
+    expect(String(process.env.CLAUDE_CODE_USE_OPENAI)).toBe('1')
+
+    // Now switch to the anthropic profile
+    const result = setActiveProviderProfile('anthro_prof')
+
+    expect(result?.id).toBe('anthro_prof')
+    expect(process.env.ANTHROPIC_MODEL).toBe('claude-sonnet-4-6')
+    expect(process.env.ANTHROPIC_BASE_URL).toBe('https://api.anthropic.com')
+    expect(process.env.CLAUDE_CODE_USE_OPENAI).toBeUndefined()
+    expect(process.env.OPENAI_MODEL).toBeUndefined()
+    expect(process.env.OPENAI_BASE_URL).toBeUndefined()
+    expect(process.env.OPENAI_API_KEY).toBeUndefined()
+    expect(process.env.CLAUDE_CODE_PROVIDER_PROFILE_ENV_APPLIED_ID).toBe(
+      'anthro_prof',
+    )
+  })
+
+  test('clears anthropic model env and sets openai model env when switching from anthropic to openai provider', async () => {
+    const { setActiveProviderProfile } =
+      await importFreshProviderProfileModules()
+    const anthropicProfile = buildProfile({
+      id: 'anthro_prof',
+      name: 'Anthropic Provider',
+      provider: 'anthropic',
+      baseUrl: 'https://api.anthropic.com',
+      model: 'claude-sonnet-4-6',
+      apiKey: 'sk-ant-key',
+    })
+    const openaiProfile = buildProfile({
+      id: 'openai_prof',
+      name: 'OpenAI Provider',
+      provider: 'openai',
+      baseUrl: 'https://api.openai.com/v1',
+      model: 'gpt-4o',
+      apiKey: 'sk-openai-key',
+    })
+
+    saveMockGlobalConfig(current => ({
+      ...current,
+      providerProfiles: [anthropicProfile, openaiProfile],
+    }))
+
+    // First activate the anthropic profile
+    setActiveProviderProfile('anthro_prof')
+    expect(process.env.ANTHROPIC_MODEL).toBe('claude-sonnet-4-6')
+    expect(process.env.ANTHROPIC_BASE_URL).toBe('https://api.anthropic.com')
+
+    // Now switch to the openai profile
+    const result = setActiveProviderProfile('openai_prof')
+
+    expect(result?.id).toBe('openai_prof')
+    expect(String(process.env.CLAUDE_CODE_USE_OPENAI)).toBe('1')
+    expect(process.env.OPENAI_MODEL).toBe('gpt-4o')
+    expect(process.env.OPENAI_BASE_URL).toBe('https://api.openai.com/v1')
+    // ANTHROPIC_MODEL is set to the profile model for all provider types
+    expect(process.env.ANTHROPIC_MODEL).toBe('gpt-4o')
+    expect(process.env.ANTHROPIC_BASE_URL).toBeUndefined()
+    expect(process.env.ANTHROPIC_API_KEY).toBeUndefined()
+    expect(process.env.CLAUDE_CODE_PROVIDER_PROFILE_ENV_APPLIED_ID).toBe(
+      'openai_prof',
+    )
+  })
+
+  test('returns null for non-existent profile id', async () => {
+    const { setActiveProviderProfile } =
+      await importFreshProviderProfileModules()
+    const openaiProfile = buildProfile({ id: 'existing_prof' })
+
+    saveMockGlobalConfig(current => ({
+      ...current,
+      providerProfiles: [openaiProfile],
+    }))
+
+    const result = setActiveProviderProfile('nonexistent_prof')
+
+    expect(result).toBeNull()
+  })
+})
+
 describe('deleteProviderProfile', () => {
  test('deleting final profile clears provider env when active profile applied it', async () => {
    const {
@@ -429,3 +625,82 @@ describe('deleteProviderProfile', () => {
    expect(process.env.OPENAI_MODEL).toBe('qwen2.5:3b')
  })
 })
+
+describe('getProfileModelOptions', () => {
+  test('generates options for multi-model profile', async () => {
+    const { getProfileModelOptions } =
+      await importFreshProviderProfileModules()
+
+    const options = getProfileModelOptions(
+      buildProfile({
+        name: 'Test Provider',
+        model: 'glm-4.7, glm-4.7-flash, glm-4.7-plus',
+      }),
+    )
+
+    expect(options).toEqual([
+      { value: 'glm-4.7', label: 'glm-4.7', description: 'Provider: Test Provider' },
+      { value: 'glm-4.7-flash', label: 'glm-4.7-flash', description: 'Provider: Test Provider' },
+      { value: 'glm-4.7-plus', label: 'glm-4.7-plus', description: 'Provider: Test Provider' },
+    ])
+  })
+
+  test('returns single option for single-model profile', async () => {
+    const { getProfileModelOptions } =
+      await importFreshProviderProfileModules()
+
+    const options = getProfileModelOptions(
+      buildProfile({
+        name: 'Single Model',
+        model: 'llama3.1:8b',
+      }),
+    )
+
+    expect(options).toEqual([
+      { value: 'llama3.1:8b', label: 'llama3.1:8b', description: 'Provider: Single Model' },
+    ])
+  })
+
+  test('returns empty array for empty model field', async () => {
+    const { getProfileModelOptions } =
+      await importFreshProviderProfileModules()
+
+    const options = getProfileModelOptions(
+      buildProfile({
+        name: 'Empty',
+        model: '',
+      }),
+    )
+
+    expect(options).toEqual([])
+  })
+})
+
+describe('setActiveProviderProfile model cache', () => {
+  test('populates model cache with all models from multi-model profile on activation', async () => {
+    const {
+      setActiveProviderProfile,
+      getActiveOpenAIModelOptionsCache,
+    } = await importFreshProviderProfileModules()
+
+    mockConfigState = {
+      ...createMockConfigState(),
+      providerProfiles: [
+        buildProfile({
+          id: 'multi_provider',
+          name: 'Multi Provider',
+          model: 'glm-4.7, glm-4.7-flash, glm-4.7-plus',
+          baseUrl: 'https://api.example.com/v1',
+        }),
+      ],
+    }
+
+    setActiveProviderProfile('multi_provider')
+
+    const cache = getActiveOpenAIModelOptionsCache()
+    const cacheValues = cache.map(opt => opt.value)
+    expect(cacheValues).toContain('glm-4.7')
+    expect(cacheValues).toContain('glm-4.7-flash')
+    expect(cacheValues).toContain('glm-4.7-plus')
+  })
+})
--- a/src/utils/providerProfiles.ts
+++ b/src/utils/providerProfiles.ts
@@ -5,6 +5,7 @@ import {
  type ProviderProfile,
 } from './config.js'
 import type { ModelOption } from './model/modelOptions.js'
+import { getPrimaryModel, parseModelList } from './providerModels.js'

 export type ProviderPreset =
  | 'anthropic'
@@ -331,7 +332,7 @@ function isProcessEnvAlignedWithProfile(
    return (
      !hasProviderSelectionFlags(processEnv) &&
      sameOptionalEnvValue(processEnv.ANTHROPIC_BASE_URL, profile.baseUrl) &&
-      sameOptionalEnvValue(processEnv.ANTHROPIC_MODEL, profile.model) &&
+      sameOptionalEnvValue(processEnv.ANTHROPIC_MODEL, getPrimaryModel(profile.model)) &&
      (!includeApiKey ||
        sameOptionalEnvValue(processEnv.ANTHROPIC_API_KEY, profile.apiKey))
    )
@@ -346,7 +347,7 @@ function isProcessEnvAlignedWithProfile(
    processEnv.CLAUDE_CODE_USE_VERTEX === undefined &&
    processEnv.CLAUDE_CODE_USE_FOUNDRY === undefined &&
    sameOptionalEnvValue(processEnv.OPENAI_BASE_URL, profile.baseUrl) &&
-    sameOptionalEnvValue(processEnv.OPENAI_MODEL, profile.model) &&
+    sameOptionalEnvValue(processEnv.OPENAI_MODEL, getPrimaryModel(profile.model)) &&
    (!includeApiKey ||
      sameOptionalEnvValue(processEnv.OPENAI_API_KEY, profile.apiKey))
  )
@@ -397,7 +398,7 @@ export function applyProviderProfileToProcessEnv(profile: ProviderProfile): void
  process.env[PROFILE_ENV_APPLIED_FLAG] = '1'
  process.env[PROFILE_ENV_APPLIED_ID] = profile.id

-  process.env.ANTHROPIC_MODEL = profile.model
+  process.env.ANTHROPIC_MODEL = getPrimaryModel(profile.model)
  if (profile.provider === 'anthropic') {
    process.env.ANTHROPIC_BASE_URL = profile.baseUrl

@@ -416,7 +417,7 @@ export function applyProviderProfileToProcessEnv(profile: ProviderProfile): void

  process.env.CLAUDE_CODE_USE_OPENAI = '1'
  process.env.OPENAI_BASE_URL = profile.baseUrl
-  process.env.OPENAI_MODEL = profile.model
+  process.env.OPENAI_MODEL = getPrimaryModel(profile.model)

  if (profile.apiKey) {
    process.env.OPENAI_API_KEY = profile.apiKey
@@ -581,6 +582,16 @@ export function persistActiveProviderProfileModel(
    return null
  }

+  // If the model is already part of the profile's model list, don't
+  // overwrite the field. This preserves comma-separated model lists like
+  // "glm-4.5, glm-4.7". Switching between models in the list is a
+  // session-level choice handled by mainLoopModelOverride, not a profile
+  // edit — the profile's model list should only change via explicit edit.
+  const existingModels = parseModelList(activeProfile.model)
+  if (existingModels.includes(nextModel)) {
+    return activeProfile
+  }
+
  saveGlobalConfig(current => {
    const currentProfiles = getProviderProfiles(current)
    const profileIndex = currentProfiles.findIndex(
@@ -623,6 +634,23 @@ export function persistActiveProviderProfileModel(
  return resolvedProfile
 }

+/**
+ * Generate model options from a provider profile's model field.
+ * Each comma-separated model becomes a separate option in the picker.
+ */
+export function getProfileModelOptions(profile: ProviderProfile): ModelOption[] {
+  const models = parseModelList(profile.model)
+  if (models.length === 0) {
+    return []
+  }
+
+  return models.map(model => ({
+    value: model,
+    label: model,
+    description: `Provider: ${profile.name}`,
+  }))
+}
+
 export function setActiveProviderProfile(
  profileId: string,
 ): ProviderProfile | null {
@@ -634,10 +662,20 @@ export function setActiveProviderProfile(
    return null
  }

+  const profileModelOptions = getProfileModelOptions(activeProfile)
+
  saveGlobalConfig(config => ({
    ...config,
    activeProviderProfileId: profileId,
-    openaiAdditionalModelOptionsCache: getModelCacheByProfile(profileId, config),
+    openaiAdditionalModelOptionsCache: profileModelOptions.length > 0
+      ? profileModelOptions
+      : getModelCacheByProfile(profileId, config),
+    openaiAdditionalModelOptionsCacheByProfile: {
+      ...(config.openaiAdditionalModelOptionsCacheByProfile ?? {}),
+      [profileId]: profileModelOptions.length > 0
+        ? profileModelOptions
+        : (config.openaiAdditionalModelOptionsCacheByProfile?.[profileId] ?? []),
+    },
  }))

  applyProviderProfileToProcessEnv(activeProfile)