diff --git a/README.md b/README.md
index dda426f..ff79986 100644
--- a/README.md
+++ b/README.md
@@ -112,11 +112,11 @@ opencode run "Hello" --model=google/antigravity-claude-opus-4-6-thinking --varia
 
 | Model | Variants | Notes |
 |-------|----------|-------|
-| `antigravity-gemini-3-pro` | low, high | Gemini 3 Pro with thinking |
 | `antigravity-gemini-3.1-pro` | low, high | Gemini 3.1 Pro with thinking (rollout-dependent) |
 | `antigravity-gemini-3-flash` | minimal, low, medium, high | Gemini 3 Flash with thinking |
 | `antigravity-claude-sonnet-4-6` | — | Claude Sonnet 4.6 |
-| `antigravity-claude-opus-4-6-thinking` | low, max | Claude Opus 4.6 with extended thinking |
+| `antigravity-claude-sonnet-4-6-thinking` | low, medium, high | Claude Sonnet 4.6 with extended thinking |
+| `antigravity-claude-opus-4-6-thinking` | low, medium, high, max | Claude Opus 4.6 with extended thinking |
 
 **Gemini CLI quota** (separate from Antigravity; used when `cli_first` is true or as fallback):
 
@@ -125,7 +125,7 @@ opencode run "Hello" --model=google/antigravity-claude-opus-4-6-thinking --varia
 | `gemini-2.5-flash` | Gemini 2.5 Flash |
 | `gemini-2.5-pro` | Gemini 2.5 Pro |
 | `gemini-3-flash-preview` | Gemini 3 Flash (preview) |
-| `gemini-3-pro-preview` | Gemini 3 Pro (preview) |
+| `gemini-3-pro-preview` | Gemini 3 Pro Preview |
 | `gemini-3.1-pro-preview` | Gemini 3.1 Pro (preview, rollout-dependent) |
 | `gemini-3.1-pro-preview-customtools` | Gemini 3.1 Pro Preview Custom Tools (preview, rollout-dependent) |
 
@@ -155,15 +155,6 @@ Add this to your `~/.config/opencode/opencode.json`:
   "provider": {
     "google": {
       "models": {
-        "antigravity-gemini-3-pro": {
-          "name": "Gemini 3 Pro (Antigravity)",
-          "limit": { "context": 1048576, "output": 65535 },
-          "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] },
-          "variants": {
-            "low": { "thinkingLevel": "low" },
-            "high": { "thinkingLevel": "high" }
-          }
-        },
         "antigravity-gemini-3.1-pro": {
           "name": "Gemini 3.1 Pro (Antigravity)",
           "limit": { "context": 1048576, "output": 65535 },
@@ -189,13 +180,25 @@ Add this to your `~/.config/opencode/opencode.json`:
           "limit": { "context": 200000, "output": 64000 },
           "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] }
         },
+        "antigravity-claude-sonnet-4-6-thinking": {
+          "name": "Claude Sonnet 4.6 Thinking (Antigravity)",
+          "limit": { "context": 200000, "output": 64000 },
+          "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] },
+          "variants": {
+            "low": { "thinkingConfig": { "thinkingBudget": 8192 } },
+            "medium": { "thinkingConfig": { "thinkingBudget": 16384 } },
+            "high": { "thinkingConfig": { "thinkingBudget": 32768 } }
+          }
+        },
         "antigravity-claude-opus-4-6-thinking": {
           "name": "Claude Opus 4.6 Thinking (Antigravity)",
           "limit": { "context": 200000, "output": 64000 },
           "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] },
           "variants": {
             "low": { "thinkingConfig": { "thinkingBudget": 8192 } },
-            "max": { "thinkingConfig": { "thinkingBudget": 32768 } }
+            "medium": { "thinkingConfig": { "thinkingBudget": 16384 } },
+            "high": { "thinkingConfig": { "thinkingBudget": 32768 } },
+            "max": { "thinkingConfig": { "thinkingBudget": 65536 } }
           }
         },
         "gemini-2.5-flash": {
@@ -422,7 +425,7 @@ If you encounter errors during a session:
 {
   "google_auth": false,
   "agents": {
-    "frontend-ui-ux-engineer": { "model": "google/antigravity-gemini-3-pro" },
+    "frontend-ui-ux-engineer": { "model": "google/antigravity-gemini-3.1-pro" },
     "document-writer": { "model": "google/antigravity-gemini-3-flash" }
   }
 }
@@ -560,7 +563,7 @@ Disable built-in auth and override agent models in `oh-my-opencode.json`:
 {
   "google_auth": false,
   "agents": {
-    "frontend-ui-ux-engineer": { "model": "google/antigravity-gemini-3-pro" },
+    "frontend-ui-ux-engineer": { "model": "google/antigravity-gemini-3.1-pro" },
     "document-writer": { "model": "google/antigravity-gemini-3-flash" },
     "multimodal-looker": { "model": "google/antigravity-gemini-3-flash" }
   }
diff --git a/docs/ANTIGRAVITY_API_SPEC.md b/docs/ANTIGRAVITY_API_SPEC.md
index 8200d63..42eaa7a 100644
--- a/docs/ANTIGRAVITY_API_SPEC.md
+++ b/docs/ANTIGRAVITY_API_SPEC.md
@@ -79,9 +79,8 @@ Accept: text/event-stream
 | Model Name | Model ID | Type | Status |
 |------------|----------|------|--------|
 | Claude Sonnet 4.6 | `claude-sonnet-4-6` | Anthropic | ✅ Verified |
+| Claude Sonnet 4.6 Thinking | `claude-sonnet-4-6-thinking` | Anthropic | ⏳ Pending Rollout |
 | Claude Opus 4.6 Thinking | `claude-opus-4-6-thinking` | Anthropic | ✅ Verified |
-| Gemini 3 Pro High | `gemini-3-pro-high` | Google | ✅ Verified |
-| Gemini 3 Pro Low | `gemini-3-pro-low` | Google | ✅ Verified |
 | GPT-OSS 120B Medium | `gpt-oss-120b-medium` | Other | ✅ Verified |
 
 ---
diff --git a/docs/MODEL-VARIANTS.md b/docs/MODEL-VARIANTS.md
index 6f3c2d6..8811cd0 100644
--- a/docs/MODEL-VARIANTS.md
+++ b/docs/MODEL-VARIANTS.md
@@ -50,8 +50,8 @@ The plugin accepts different variant formats depending on the model family:
 
 Gemini 3 models use string-based thinking levels. Available levels differ by model:
 
-| Level | Flash | Pro | Description |
-|-------|-------|-----|-------------|
+| Level | Flash | Pro (3.1) | Description |
+|-------|-------|-------------|-------------|
 | `minimal` | ✅ | ❌ | Minimal thinking, lowest latency |
 | `low` | ✅ | ✅ | Light thinking |
 | `medium` | ✅ | ❌ | Balanced thinking |
@@ -59,22 +59,6 @@ Gemini 3 models use string-based thinking levels. Available levels differ by mod
 
 > **Note:** The API rejects invalid levels (e.g., `"minimal"` on Pro). Configure variants accordingly.
 
-### Gemini 3 Pro Example
-
-```json
-{
-  "antigravity-gemini-3-pro": {
-    "name": "Gemini 3 Pro (Antigravity)",
-    "limit": { "context": 1048576, "output": 65535 },
-    "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] },
-    "variants": {
-      "low": { "thinkingLevel": "low" },
-      "high": { "thinkingLevel": "high" }
-    }
-  }
-}
-```
-
 ### Gemini 3 Flash Example
 
 ```json
@@ -99,22 +83,34 @@ Gemini 3 models use string-based thinking levels. Available levels differ by mod
 
 Claude models use token-based thinking budgets:
 
-| Variant | Budget | Description |
-|---------|--------|-------------|
-| `low` | 8192 | Light thinking |
-| `max` | 32768 | Maximum thinking |
+| Model | Variants | Budgets |
+|-------|----------|---------|
+| `antigravity-claude-opus-4-6-thinking` | `low`, `medium`, `high`, `max` | 8192, 16384, 32768, 65536 |
+| `antigravity-claude-sonnet-4-6-thinking` | `low`, `medium`, `high` | 8192, 16384, 32768 |
 
 ### Claude Example
 
 ```json
 {
+  "antigravity-claude-sonnet-4-6-thinking": {
+    "name": "Claude Sonnet 4.6 Thinking (Antigravity)",
+    "limit": { "context": 200000, "output": 64000 },
+    "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] },
+    "variants": {
+      "low": { "thinkingConfig": { "thinkingBudget": 8192 } },
+      "medium": { "thinkingConfig": { "thinkingBudget": 16384 } },
+      "high": { "thinkingConfig": { "thinkingBudget": 32768 } }
+    }
+  },
   "antigravity-claude-opus-4-6-thinking": {
     "name": "Claude Opus 4.6 Thinking (Antigravity)",
     "limit": { "context": 200000, "output": 64000 },
     "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] },
     "variants": {
       "low": { "thinkingConfig": { "thinkingBudget": 8192 } },
-      "max": { "thinkingConfig": { "thinkingBudget": 32768 } }
+      "medium": { "thinkingConfig": { "thinkingBudget": 16384 } },
+      "high": { "thinkingConfig": { "thinkingBudget": 32768 } },
+      "max": { "thinkingConfig": { "thinkingBudget": 65536 } }
     }
   }
 }
@@ -157,9 +153,10 @@ Tier-suffixed model names are still accepted:
 - `antigravity-claude-opus-4-6-thinking-low`
 - `antigravity-claude-opus-4-6-thinking-medium`
 - `antigravity-claude-opus-4-6-thinking-high`
-- `antigravity-gemini-3-pro-low`
-- `antigravity-gemini-3-pro-high`
-- `gemini-3-pro-low`
+- `antigravity-claude-opus-4-6-thinking-max`
+- `antigravity-claude-sonnet-4-6-thinking-low`
+- `antigravity-claude-sonnet-4-6-thinking-medium`
+- `antigravity-claude-sonnet-4-6-thinking-high`
 - `gemini-3-flash-medium`
 
 However, **we recommend using simplified model names with variants** for:
diff --git a/docs/TROUBLESHOOTING.md b/docs/TROUBLESHOOTING.md
index 9482b2e..fd195c8 100644
--- a/docs/TROUBLESHOOTING.md
+++ b/docs/TROUBLESHOOTING.md
@@ -431,7 +431,7 @@ v1.2.7+ uses explicit `antigravity-` prefix:
 
 | Old Name | New Name |
 |----------|----------|
-| `gemini-3-pro-low` | `antigravity-gemini-3-pro` |
+| `gemini-3.1-pro-low` | `antigravity-gemini-3.1-pro` |
 | `claude-sonnet-4-6` | `antigravity-claude-sonnet-4-6` |
 
 Use the `antigravity-` prefixed model names shown above.
diff --git a/script/test-cross-model-e2e.sh b/script/test-cross-model-e2e.sh
index 073c74d..f76dcc6 100755
--- a/script/test-cross-model-e2e.sh
+++ b/script/test-cross-model-e2e.sh
@@ -3,7 +3,7 @@
 # Tests fix for "Invalid `signature` in `thinking` block" error
 #
 # Models tested:
-# 1. Gemini (google/antigravity-gemini-3-pro-low, gemini-3-flash)
+# 1. Gemini (google/antigravity-gemini-3.1-pro-low, gemini-3-flash)
 # 2. Claude via Anthropic (anthropic/claude-opus-4-5)
 # 3. Claude via Google (google/antigravity-claude-*-thinking-*)
 # 4. OpenAI (openai/gpt-5.2-medium)
@@ -42,7 +42,7 @@ echo ""
 # Test 1: Gemini → Anthropic Claude (original bug + direct Anthropic API)
 echo "Test 1: Gemini Pro → Anthropic Claude Opus (direct API)"
 log_info "Step 1: Gemini with thinking + tool..."
-opencode run -m google/antigravity-gemini-3-pro-low \
+opencode run -m google/antigravity-gemini-3.1-pro-low \
   "Run: echo 'Test1-Gemini'. Think about sequences." \
   > /tmp/e2e-t1-s1.log 2>&1 || true
 
@@ -67,7 +67,7 @@ echo ""
 # Test 2: Gemini → Google Claude (Google-hosted Claude)
 echo "Test 2: Gemini Pro → Google Claude Opus Thinking"
 log_info "Step 1: Gemini with thinking + tool..."
-opencode run -m google/antigravity-gemini-3-pro-low \
+opencode run -m google/antigravity-gemini-3.1-pro-low \
   "Run: echo 'Test2-Gemini'. Think about this." \
   > /tmp/e2e-t2-s1.log 2>&1 || true
 
@@ -92,7 +92,7 @@ echo ""
 # Test 3: Gemini → OpenAI
 echo "Test 3: Gemini Pro → OpenAI GPT-5.2"
 log_info "Step 1: Gemini with thinking + tool..."
-opencode run -m google/antigravity-gemini-3-pro-low \
+opencode run -m google/antigravity-gemini-3.1-pro-low \
   "Run: echo 'Test3-Gemini'. Think about AI models." \
   > /tmp/e2e-t3-s1.log 2>&1 || true
 
@@ -129,7 +129,7 @@ if [ -z "$SID" ]; then
 else
   log_info "Session: $SID"
   log_info "Step 2: Gemini + thinking + tool..."
-  opencode run -s "$SID" -m google/antigravity-gemini-3-pro-low \
+  opencode run -s "$SID" -m google/antigravity-gemini-3.1-pro-low \
     "Run: echo 'Test4-Gemini'. Think about reversal." \
     > /tmp/e2e-t4-s2.log 2>&1 || true
   
@@ -173,7 +173,7 @@ echo ""
 # Test 6: 5-Model Round-Robin (all models in sequence)
 echo "Test 6: 5-Model Round-Robin"
 log_info "Turn 1: Gemini Pro Low..."
-opencode run -m google/antigravity-gemini-3-pro-low \
+opencode run -m google/antigravity-gemini-3.1-pro-low \
   "Run: echo 'Turn1'. Think about the chain." \
   > /tmp/e2e-t6-s1.log 2>&1 || true
 
diff --git a/script/test-cross-model.ts b/script/test-cross-model.ts
index 6c13731..4452541 100644
--- a/script/test-cross-model.ts
+++ b/script/test-cross-model.ts
@@ -58,7 +58,7 @@ function runTests(): void {
   let failed = 0;
 
   console.log('Test 1: Model family detection');
-  const geminiFamily = getModelFamily('gemini-3-pro-low');
+  const geminiFamily = getModelFamily('gemini-3.1-pro-low');
   const claudeFamily = getModelFamily('claude-opus-4-6-thinking-medium');
   if (geminiFamily === 'gemini' && claudeFamily === 'claude') {
     console.log('  ✅ PASS: Model families detected correctly');
diff --git a/script/test-gemini-cli-e2e.sh b/script/test-gemini-cli-e2e.sh
index 6cc4664..572e21c 100755
--- a/script/test-gemini-cli-e2e.sh
+++ b/script/test-gemini-cli-e2e.sh
@@ -5,8 +5,9 @@
 # Models tested:
 # 1. google/gemini-2.5-pro
 # 2. google/gemini-2.5-flash
-# 3. google/gemini-3-pro-preview
-# 4. google/gemini-3-flash-preview
+# 3. google/gemini-3-flash-preview
+# 4. google/gemini-3-pro-preview
+# 5. google/gemini-3.1-pro-preview
 
 set -euo pipefail
 
@@ -98,8 +99,12 @@ echo "Test 4: google/gemini-3-pro-preview"
 test_model "google/gemini-3-pro-preview" "gemini-3-pro-preview" || true
 echo ""
 
-# Test 5: Cross-model session (gemini-cli → antigravity)
-echo "Test 5: Cross-model session (gemini-cli → antigravity-gemini)"
+echo "Test 5: google/gemini-3.1-pro-preview"
+test_model "google/gemini-3.1-pro-preview" "gemini-3.1-pro-preview" || true
+echo ""
+
+# Test 6: Cross-model session (gemini-cli → antigravity)
+echo "Test 6: Cross-model session (gemini-cli → antigravity-gemini)"
 log_info "Step 1: Start with gemini-2.5-flash..."
 timeout 60 opencode run -m google/gemini-2.5-flash \
   "Say: SESSION_START" \
@@ -110,7 +115,7 @@ sleep 1
 SID=$(opencode session list 2>/dev/null | grep -oP 'ses_[a-zA-Z0-9]+' | head -1 || true)
 
 if [ -z "$SID" ]; then
-  log_fail "Test 5 - No session ID created"
+  log_fail "Test 6 - No session ID created"
 else
   log_info "Session: $SID"
   log_info "Step 2: Switch to antigravity-gemini-3-flash..."
@@ -119,17 +124,17 @@ else
     2>&1 > /tmp/gemini-cli-e2e-cross-s2.log || true
   
   if check_auth_error /tmp/gemini-cli-e2e-cross-s2.log; then
-    log_fail "Test 5 - Auth error on cross-model switch"
+    log_fail "Test 6 - Auth error on cross-model switch"
   else
-    log_pass "Test 5 - Cross-model session (gemini-cli → antigravity)"
+    log_pass "Test 6 - Cross-model session (gemini-cli → antigravity)"
   fi
 fi
 echo ""
 
-# Test 6: Reverse cross-model (antigravity → gemini-cli)
-echo "Test 6: Cross-model session (antigravity → gemini-cli)"
-log_info "Step 1: Start with antigravity-gemini-3-pro-low..."
-timeout 60 opencode run -m google/antigravity-gemini-3-pro-low \
+# Test 7: Reverse cross-model (antigravity → gemini-cli)
+echo "Test 7: Cross-model session (antigravity → gemini-cli)"
+log_info "Step 1: Start with antigravity-gemini-3.1-pro-low..."
+timeout 60 opencode run -m google/antigravity-gemini-3.1-pro-low \
   "Say: ANTIGRAVITY_START" \
   2>&1 > /tmp/gemini-cli-e2e-reverse-s1.log || true
 
@@ -137,7 +142,7 @@ sleep 1
 SID=$(opencode session list 2>/dev/null | grep -oP 'ses_[a-zA-Z0-9]+' | head -1 || true)
 
 if [ -z "$SID" ]; then
-  log_fail "Test 6 - No session ID created"
+  log_fail "Test 7 - No session ID created"
 else
   log_info "Session: $SID"
   log_info "Step 2: Switch to gemini-2.5-pro..."
@@ -146,9 +151,9 @@ else
     2>&1 > /tmp/gemini-cli-e2e-reverse-s2.log || true
   
   if check_auth_error /tmp/gemini-cli-e2e-reverse-s2.log; then
-    log_fail "Test 6 - Auth error on reverse cross-model switch"
+    log_fail "Test 7 - Auth error on reverse cross-model switch"
   else
-    log_pass "Test 6 - Cross-model session (antigravity → gemini-cli)"
+    log_pass "Test 7 - Cross-model session (antigravity → gemini-cli)"
   fi
 fi
 echo ""
diff --git a/script/test-models.ts b/script/test-models.ts
index 5ce9b87..4d1fd9c 100644
--- a/script/test-models.ts
+++ b/script/test-models.ts
@@ -10,19 +10,24 @@ const MODELS: ModelTest[] = [
   // Gemini CLI (direct Google API)
   { model: "google/gemini-3-flash-preview", category: "gemini-cli" },
   { model: "google/gemini-3-pro-preview", category: "gemini-cli" },
+  { model: "google/gemini-3.1-pro-preview", category: "gemini-cli" },
   { model: "google/gemini-2.5-pro", category: "gemini-cli" },
   { model: "google/gemini-2.5-flash", category: "gemini-cli" },
 
   // Antigravity Gemini
-  { model: "google/antigravity-gemini-3-pro-low", category: "antigravity-gemini" },
-  { model: "google/antigravity-gemini-3-pro-high", category: "antigravity-gemini" },
+  { model: "google/antigravity-gemini-3.1-pro-low", category: "antigravity-gemini" },
+  { model: "google/antigravity-gemini-3.1-pro-high", category: "antigravity-gemini" },
   { model: "google/antigravity-gemini-3-flash", category: "antigravity-gemini" },
 
   // Antigravity Claude
   { model: "google/antigravity-claude-sonnet-4-6", category: "antigravity-claude" },
+  { model: "google/antigravity-claude-sonnet-4-6-thinking-low", category: "antigravity-claude" },
+  { model: "google/antigravity-claude-sonnet-4-6-thinking-medium", category: "antigravity-claude" },
+  { model: "google/antigravity-claude-sonnet-4-6-thinking-high", category: "antigravity-claude" },
   { model: "google/antigravity-claude-opus-4-6-thinking-low", category: "antigravity-claude" },
   { model: "google/antigravity-claude-opus-4-6-thinking-medium", category: "antigravity-claude" },
   { model: "google/antigravity-claude-opus-4-6-thinking-high", category: "antigravity-claude" },
+  { model: "google/antigravity-claude-opus-4-6-thinking-max", category: "antigravity-claude" },
 ];
 
 const TEST_PROMPT = "Reply with exactly one word: WORKING";
diff --git a/scripts/setup-opencode-pi.sh b/scripts/setup-opencode-pi.sh
index 1c6919c..9cf127c 100755
--- a/scripts/setup-opencode-pi.sh
+++ b/scripts/setup-opencode-pi.sh
@@ -32,8 +32,8 @@ if [ ! -f "$CONFIG_FILE" ]; then
   "provider": {
     "google": {
       "models": {
-        "antigravity-gemini-3-pro": {
-          "name": "Gemini 3 Pro (Antigravity)",
+        "antigravity-gemini-3.1-pro": {
+          "name": "Gemini 3.1 Pro (Antigravity)",
           "limit": { "context": 1048576, "output": 65535 },
           "modalities": { "input": ["text", "image", "pdf"], "output": ["text"] },
           "variants": {
diff --git a/src/plugin/accounts.test.ts b/src/plugin/accounts.test.ts
index ad0b62b..b898d59 100644
--- a/src/plugin/accounts.test.ts
+++ b/src/plugin/accounts.test.ts
@@ -1863,6 +1863,6 @@ describe("resolveQuotaGroup", () => {
   it("model takes precedence over family", () => {
     // Even if family says claude, model determines the quota group
     expect(resolveQuotaGroup("gemini", "gemini-2.5-flash")).toBe("gemini-flash");
-    expect(resolveQuotaGroup("gemini", "gemini-3-pro")).toBe("gemini-pro");
+    expect(resolveQuotaGroup("gemini", "gemini-3.1-pro")).toBe("gemini-pro");
   });
 });
diff --git a/src/plugin/antigravity-first-fallback.test.ts b/src/plugin/antigravity-first-fallback.test.ts
index 917872e..a22ff23 100644
--- a/src/plugin/antigravity-first-fallback.test.ts
+++ b/src/plugin/antigravity-first-fallback.test.ts
@@ -138,14 +138,14 @@ describe("Antigravity-first fallback", () => {
       const manager = new AccountManager(undefined, stored);
       const accounts = manager.getAccounts();
       
-      // Mark account 0's antigravity as rate-limited for gemini-3-pro
-      manager.markRateLimited(accounts[0]!, 60000, "gemini", "antigravity", "gemini-3-pro");
+      // Mark account 0's antigravity as rate-limited for gemini-3.1-pro
+      manager.markRateLimited(accounts[0]!, 60000, "gemini", "antigravity", "gemini-3.1-pro");
 
-      // Account 1 should have antigravity available for gemini-3-pro
+      // Account 1 should have antigravity available for gemini-3.1-pro
       const hasOther = manager.hasOtherAccountWithAntigravityAvailable(
         accounts[0]!.index,
         "gemini",
-        "gemini-3-pro"
+        "gemini-3.1-pro"
       );
 
       expect(hasOther).toBe(true);
diff --git a/src/plugin/config/models.test.ts b/src/plugin/config/models.test.ts
index fd5aecc..12ec5ef 100644
--- a/src/plugin/config/models.test.ts
+++ b/src/plugin/config/models.test.ts
@@ -17,8 +17,8 @@ describe("OPENCODE_MODEL_DEFINITIONS", () => {
     expect(modelNames).toEqual([
       "antigravity-claude-opus-4-6-thinking",
       "antigravity-claude-sonnet-4-6",
+      "antigravity-claude-sonnet-4-6-thinking",
       "antigravity-gemini-3-flash",
-      "antigravity-gemini-3-pro",
       "antigravity-gemini-3.1-pro",
       "gemini-2.5-flash",
       "gemini-2.5-pro",
@@ -30,11 +30,6 @@ describe("OPENCODE_MODEL_DEFINITIONS", () => {
   });
 
   it("defines Gemini 3 variants for Antigravity models", () => {
-    expect(getModel("antigravity-gemini-3-pro").variants).toEqual({
-      low: { thinkingLevel: "low" },
-      high: { thinkingLevel: "high" },
-    });
-
     expect(getModel("antigravity-gemini-3.1-pro").variants).toEqual({
       low: { thinkingLevel: "low" },
       high: { thinkingLevel: "high" },
@@ -51,7 +46,15 @@ describe("OPENCODE_MODEL_DEFINITIONS", () => {
   it("defines thinking budget variants for Claude thinking models", () => {
     expect(getModel("antigravity-claude-opus-4-6-thinking").variants).toEqual({
       low: { thinkingConfig: { thinkingBudget: 8192 } },
-      max: { thinkingConfig: { thinkingBudget: 32768 } },
+      medium: { thinkingConfig: { thinkingBudget: 16384 } },
+      high: { thinkingConfig: { thinkingBudget: 32768 } },
+      max: { thinkingConfig: { thinkingBudget: 65536 } },
+    });
+
+    expect(getModel("antigravity-claude-sonnet-4-6-thinking").variants).toEqual({
+      low: { thinkingConfig: { thinkingBudget: 8192 } },
+      medium: { thinkingConfig: { thinkingBudget: 16384 } },
+      high: { thinkingConfig: { thinkingBudget: 32768 } },
     });
   });
 });
diff --git a/src/plugin/config/models.ts b/src/plugin/config/models.ts
index 641d2e1..f8487a4 100644
--- a/src/plugin/config/models.ts
+++ b/src/plugin/config/models.ts
@@ -38,15 +38,6 @@ const DEFAULT_MODALITIES: ModelModalities = {
 };
 
 export const OPENCODE_MODEL_DEFINITIONS: OpencodeModelDefinitions = {
-  "antigravity-gemini-3-pro": {
-    name: "Gemini 3 Pro (Antigravity)",
-    limit: { context: 1048576, output: 65535 },
-    modalities: DEFAULT_MODALITIES,
-    variants: {
-      low: { thinkingLevel: "low" },
-      high: { thinkingLevel: "high" },
-    },
-  },
   "antigravity-gemini-3.1-pro": {
     name: "Gemini 3.1 Pro (Antigravity)",
     limit: { context: 1048576, output: 65535 },
@@ -72,13 +63,25 @@ export const OPENCODE_MODEL_DEFINITIONS: OpencodeModelDefinitions = {
     limit: { context: 200000, output: 64000 },
     modalities: DEFAULT_MODALITIES,
   },
+  "antigravity-claude-sonnet-4-6-thinking": {
+    name: "Claude Sonnet 4.6 Thinking (Antigravity)",
+    limit: { context: 200000, output: 64000 },
+    modalities: DEFAULT_MODALITIES,
+    variants: {
+      low: { thinkingConfig: { thinkingBudget: 8192 } },
+      medium: { thinkingConfig: { thinkingBudget: 16384 } },
+      high: { thinkingConfig: { thinkingBudget: 32768 } },
+    },
+  },
   "antigravity-claude-opus-4-6-thinking": {
     name: "Claude Opus 4.6 Thinking (Antigravity)",
     limit: { context: 200000, output: 64000 },
     modalities: DEFAULT_MODALITIES,
     variants: {
       low: { thinkingConfig: { thinkingBudget: 8192 } },
-      max: { thinkingConfig: { thinkingBudget: 32768 } },
+      medium: { thinkingConfig: { thinkingBudget: 16384 } },
+      high: { thinkingConfig: { thinkingBudget: 32768 } },
+      max: { thinkingConfig: { thinkingBudget: 65536 } },
     },
   },
   "gemini-2.5-flash": {
diff --git a/src/plugin/config/updater.test.ts b/src/plugin/config/updater.test.ts
index 83634ed..7708a56 100644
--- a/src/plugin/config/updater.test.ts
+++ b/src/plugin/config/updater.test.ts
@@ -66,7 +66,6 @@ describe("updateOpencodeConfig", () => {
     // Old model should be replaced
     expect(writtenConfig.provider.google.models["old-model"]).toBeUndefined();
     // New models should be present
-    expect(writtenConfig.provider.google.models["antigravity-gemini-3-pro"]).toBeDefined();
     expect(writtenConfig.provider.google.models["antigravity-claude-sonnet-4-6"]).toBeDefined();
   });
 
@@ -235,7 +234,6 @@ describe("updateOpencodeConfig", () => {
     expect(writtenConfig.plugin).toContain("other-plugin");
     expect(writtenConfig.plugin).toContain("opencode-antigravity-auth@latest");
     expect(writtenConfig.provider.google.region).toBe("us-central1");
-    expect(writtenConfig.provider.google.models["antigravity-gemini-3-pro"]).toBeDefined();
   });
 
   test("prefers existing opencode.jsonc when using default config path", async () => {
diff --git a/src/plugin/cross-model-integration.test.ts b/src/plugin/cross-model-integration.test.ts
index 1dc4567..3478c00 100644
--- a/src/plugin/cross-model-integration.test.ts
+++ b/src/plugin/cross-model-integration.test.ts
@@ -242,7 +242,7 @@ describe("Cross-Model Session Integration", () => {
       };
 
       const result = sanitizeCrossModelPayload(payload, {
-        targetModel: "gemini-3-pro-low",
+        targetModel: "gemini-3.1-pro-low",
       });
 
       const sanitized = result.payload as typeof payload;
@@ -352,10 +352,10 @@ describe("Cross-Model Session Integration", () => {
 
   describe("Model family detection", () => {
     it("correctly identifies Gemini models", () => {
-      expect(getModelFamily("gemini-3-pro-low")).toBe("gemini");
+      expect(getModelFamily("gemini-3.1-pro-low")).toBe("gemini");
       expect(getModelFamily("gemini-3-flash")).toBe("gemini");
       expect(getModelFamily("gemini-2.5-pro")).toBe("gemini");
-      expect(getModelFamily("gemini-3-pro-high")).toBe("gemini");
+      expect(getModelFamily("gemini-3.1-pro-high")).toBe("gemini");
     });
 
     it("correctly identifies Claude models", () => {
diff --git a/src/plugin/request-helpers.test.ts b/src/plugin/request-helpers.test.ts
index 60ecb0e..a8d8db4 100644
--- a/src/plugin/request-helpers.test.ts
+++ b/src/plugin/request-helpers.test.ts
@@ -171,7 +171,7 @@ describe("isThinkingCapableModel", () => {
   });
 
   it("returns true for models with 'gemini-3' in name", () => {
-    expect(isThinkingCapableModel("gemini-3-pro")).toBe(true);
+    expect(isThinkingCapableModel("gemini-3.1-pro")).toBe(true);
     expect(isThinkingCapableModel("GEMINI-3-flash")).toBe(true);
     expect(isThinkingCapableModel("gemini-3")).toBe(true);
   });
diff --git a/src/plugin/transform/claude.test.ts b/src/plugin/transform/claude.test.ts
index e3abef7..2519296 100644
--- a/src/plugin/transform/claude.test.ts
+++ b/src/plugin/transform/claude.test.ts
@@ -33,7 +33,7 @@ describe("isClaudeModel", () => {
   });
 
   it("returns false for non-claude models", () => {
-    expect(isClaudeModel("gemini-3-pro")).toBe(false);
+    expect(isClaudeModel("gemini-3.1-pro")).toBe(false);
     expect(isClaudeModel("gpt-4")).toBe(false);
     expect(isClaudeModel("llama-3")).toBe(false);
     expect(isClaudeModel("")).toBe(false);
@@ -60,6 +60,7 @@ describe("isClaudeThinkingModel", () => {
 
   it("returns true for prefixed thinking models", () => {
     expect(isClaudeThinkingModel("antigravity-claude-sonnet-4-5-thinking")).toBe(true);
+    expect(isClaudeThinkingModel("antigravity-claude-sonnet-4-6-thinking")).toBe(true);
     expect(isClaudeThinkingModel("google/claude-opus-4-5-thinking-high")).toBe(true);
   });
 
@@ -70,7 +71,7 @@ describe("isClaudeThinkingModel", () => {
   });
 
   it("returns false for non-claude models", () => {
-    expect(isClaudeThinkingModel("gemini-3-pro-thinking")).toBe(false);
+    expect(isClaudeThinkingModel("gemini-3.1-pro-thinking")).toBe(false);
     expect(isClaudeThinkingModel("gpt-4-thinking")).toBe(false);
   });
 
diff --git a/src/plugin/transform/cross-model-sanitizer.test.ts b/src/plugin/transform/cross-model-sanitizer.test.ts
index 7be1f32..4de3362 100644
--- a/src/plugin/transform/cross-model-sanitizer.test.ts
+++ b/src/plugin/transform/cross-model-sanitizer.test.ts
@@ -13,11 +13,12 @@ describe("cross-model-sanitizer", () => {
     it("identifies Claude models", () => {
       expect(getModelFamily("claude-opus-4-6-thinking-medium")).toBe("claude");
       expect(getModelFamily("claude-sonnet-4-6")).toBe("claude");
+      expect(getModelFamily("claude-sonnet-4-6-thinking")).toBe("claude");
       expect(getModelFamily("claude-opus-4-6-thinking-low")).toBe("claude");
     });
 
     it("identifies Gemini models", () => {
-      expect(getModelFamily("gemini-3-pro-low")).toBe("gemini");
+      expect(getModelFamily("gemini-3.1-pro-low")).toBe("gemini");
       expect(getModelFamily("gemini-3-flash")).toBe("gemini");
       expect(getModelFamily("gemini-2.5-pro")).toBe("gemini");
     });
@@ -328,7 +329,7 @@ describe("cross-model-sanitizer", () => {
       };
 
       const result = sanitizeCrossModelPayload(payload, {
-        targetModel: "gemini-3-pro-low",
+        targetModel: "gemini-3.1-pro-low",
       });
 
       expect(result.modified).toBe(true);
diff --git a/src/plugin/transform/gemini.test.ts b/src/plugin/transform/gemini.test.ts
index 4479444..070b151 100644
--- a/src/plugin/transform/gemini.test.ts
+++ b/src/plugin/transform/gemini.test.ts
@@ -28,8 +28,8 @@ describe("transform/gemini", () => {
       expect(isGeminiModel("gemini-2.5-flash")).toBe(true);
     });
 
-    it("returns true for gemini-3-pro-high", () => {
-      expect(isGeminiModel("gemini-3-pro-high")).toBe(true);
+    it("returns true for gemini-3.1-pro-high", () => {
+      expect(isGeminiModel("gemini-3.1-pro-high")).toBe(true);
     });
 
     it("returns true for uppercase GEMINI-PRO", () => {
@@ -62,12 +62,12 @@ describe("transform/gemini", () => {
   });
 
   describe("isGemini3Model", () => {
-    it("returns true for gemini-3-pro", () => {
-      expect(isGemini3Model("gemini-3-pro")).toBe(true);
+    it("returns true for gemini-3.1-pro", () => {
+      expect(isGemini3Model("gemini-3.1-pro")).toBe(true);
     });
 
-    it("returns true for gemini-3-pro-high", () => {
-      expect(isGemini3Model("gemini-3-pro-high")).toBe(true);
+    it("returns true for gemini-3.1-pro-high", () => {
+      expect(isGemini3Model("gemini-3.1-pro-high")).toBe(true);
     });
 
     it("returns true for gemini-3-flash", () => {
@@ -78,8 +78,8 @@ describe("transform/gemini", () => {
       expect(isGemini3Model("gemini-3.1-pro")).toBe(true);
     });
 
-    it("returns true for uppercase GEMINI-3-PRO", () => {
-      expect(isGemini3Model("GEMINI-3-PRO")).toBe(true);
+    it("returns true for uppercase GEMINI-3.1-PRO", () => {
+      expect(isGemini3Model("GEMINI-3.1-PRO")).toBe(true);
     });
 
     it("returns false for gemini-2.5-pro", () => {
@@ -116,8 +116,8 @@ describe("transform/gemini", () => {
       expect(isGemini25Model("GEMINI-2.5-PRO")).toBe(true);
     });
 
-    it("returns false for gemini-3-pro", () => {
-      expect(isGemini25Model("gemini-3-pro")).toBe(false);
+    it("returns false for gemini-3.1-pro", () => {
+      expect(isGemini25Model("gemini-3.1-pro")).toBe(false);
     });
 
     it("returns false for gemini-2.0-flash", () => {
@@ -376,7 +376,7 @@ describe("transform/gemini", () => {
     it("applies Gemini 3 thinking config with thinkingLevel", () => {
       const payload: RequestPayload = { contents: [] };
       applyGeminiTransforms(payload, {
-        model: "gemini-3-pro-high",
+        model: "gemini-3.1-pro-high",
         tierThinkingLevel: "high",
         normalizedThinking: { includeThoughts: true },
       });
@@ -425,7 +425,7 @@ describe("transform/gemini", () => {
     it("does not apply thinking config when normalizedThinking is undefined", () => {
       const payload: RequestPayload = { contents: [] };
       applyGeminiTransforms(payload, {
-        model: "gemini-3-pro",
+        model: "gemini-3.1-pro",
       });
       expect(payload.generationConfig).toBeUndefined();
     });
@@ -436,7 +436,7 @@ describe("transform/gemini", () => {
         generationConfig: { temperature: 0.7, maxOutputTokens: 1000 },
       };
       applyGeminiTransforms(payload, {
-        model: "gemini-3-pro-medium",
+        model: "gemini-3.1-pro-medium",
         tierThinkingLevel: "medium",
         normalizedThinking: { includeThoughts: true },
       });
@@ -464,7 +464,7 @@ describe("transform/gemini", () => {
     it("defaults includeThoughts to true when not specified", () => {
       const payload: RequestPayload = { contents: [] };
       applyGeminiTransforms(payload, {
-        model: "gemini-3-pro-low",
+        model: "gemini-3.1-pro-low",
         tierThinkingLevel: "low",
         normalizedThinking: {},
       });
@@ -475,7 +475,7 @@ describe("transform/gemini", () => {
     it("respects includeThoughts false", () => {
       const payload: RequestPayload = { contents: [] };
       applyGeminiTransforms(payload, {
-        model: "gemini-3-pro-high",
+        model: "gemini-3.1-pro-high",
         tierThinkingLevel: "high",
         normalizedThinking: { includeThoughts: false },
       });
@@ -499,7 +499,7 @@ describe("transform/gemini", () => {
       it("injects googleSearch tool when mode is 'auto'", () => {
         const payload: RequestPayload = { contents: [], tools: [] };
         applyGeminiTransforms(payload, {
-          model: "gemini-3-pro",
+          model: "gemini-3.1-pro",
           googleSearch: { mode: "auto" },
         });
         const tools = payload.tools as unknown[];
@@ -524,7 +524,7 @@ describe("transform/gemini", () => {
       it("works without threshold specified", () => {
         const payload: RequestPayload = { contents: [] };
         applyGeminiTransforms(payload, {
-          model: "gemini-3-pro",
+          model: "gemini-3.1-pro",
           googleSearch: { mode: "auto" },
         });
         const tools = payload.tools as unknown[];
@@ -535,7 +535,7 @@ describe("transform/gemini", () => {
       it("does not inject search tool when mode is 'off'", () => {
         const payload: RequestPayload = { contents: [], tools: [] };
         applyGeminiTransforms(payload, {
-          model: "gemini-3-pro",
+          model: "gemini-3.1-pro",
           googleSearch: { mode: "off" },
         });
         const tools = payload.tools as unknown[];
@@ -545,7 +545,7 @@ describe("transform/gemini", () => {
       it("does not inject search tool when googleSearch is undefined", () => {
         const payload: RequestPayload = { contents: [], tools: [] };
         applyGeminiTransforms(payload, {
-          model: "gemini-3-pro",
+          model: "gemini-3.1-pro",
         });
         const tools = payload.tools as unknown[];
         expect(tools).toHaveLength(0);
@@ -559,7 +559,7 @@ describe("transform/gemini", () => {
           ],
         };
         applyGeminiTransforms(payload, {
-          model: "gemini-3-pro",
+          model: "gemini-3.1-pro",
           googleSearch: { mode: "auto" },
         });
         const tools = payload.tools as unknown[];
@@ -571,7 +571,7 @@ describe("transform/gemini", () => {
       it("search tool is not normalized (skipped by normalizeGeminiTools)", () => {
         const payload: RequestPayload = { contents: [] };
         applyGeminiTransforms(payload, {
-          model: "gemini-3-pro",
+          model: "gemini-3.1-pro",
           googleSearch: { mode: "auto" },
         });
         const tools = payload.tools as unknown[];
@@ -604,16 +604,17 @@ describe("transform/gemini", () => {
       expect(isImageGenerationModel("GEMINI-3-PRO-IMAGE")).toBe(true);
     });
 
-    it("returns false for gemini-3-pro", () => {
-      expect(isImageGenerationModel("gemini-3-pro")).toBe(false);
+    it("returns false for gemini-3.1-pro", () => {
+      expect(isImageGenerationModel("gemini-3.1-pro")).toBe(false);
     });
 
     it("returns false for gemini-2.5-flash", () => {
       expect(isImageGenerationModel("gemini-2.5-flash")).toBe(false);
     });
 
-    it("returns false for claude-sonnet-4-6", () => {
+    it("returns false for claude-sonnet-4-6 and thinking variants", () => {
       expect(isImageGenerationModel("claude-sonnet-4-6")).toBe(false);
+      expect(isImageGenerationModel("claude-sonnet-4-6-thinking")).toBe(false);
     });
   });
 
@@ -1441,7 +1442,7 @@ describe("transform/gemini", () => {
         ],
       };
       
-      applyGeminiTransforms(payload, { model: "gemini-3-pro" });
+      applyGeminiTransforms(payload, { model: "gemini-3.1-pro" });
       
       const tools = payload.tools as Array<Record<string, unknown>>;
       expect(tools).toHaveLength(1);
@@ -1467,7 +1468,7 @@ describe("transform/gemini", () => {
       };
 
       applyGeminiTransforms(payload, {
-        model: "gemini-3-pro",
+        model: "gemini-3.1-pro",
         googleSearch: { mode: "auto" },
       });
       
diff --git a/src/plugin/transform/model-resolver.test.ts b/src/plugin/transform/model-resolver.test.ts
index e0b9b48..3c48b76 100644
--- a/src/plugin/transform/model-resolver.test.ts
+++ b/src/plugin/transform/model-resolver.test.ts
@@ -27,12 +27,12 @@ describe("resolveModelWithTier", () => {
   });
 
   describe("Gemini 3 preview models (Issue #115)", () => {
-    it("gemini-3-pro-preview gets default thinkingLevel 'low' with antigravity quota", () => {
+    it("gemini-3-pro-preview resolves as Gemini CLI-first with default thinkingLevel", () => {
       const result = resolveModelWithTier("gemini-3-pro-preview");
       expect(result.actualModel).toBe("gemini-3-pro-preview");
       expect(result.thinkingLevel).toBe("low");
-      // All Gemini models now default to antigravity
-      expect(result.quotaPreference).toBe("antigravity");
+      expect(result.quotaPreference).toBe("gemini-cli");
+      expect(result.thinkingBudget).toBeUndefined();
     });
 
     it("gemini-3.1-pro-preview gets default thinkingLevel 'low' with antigravity quota", () => {
@@ -78,12 +78,6 @@ describe("resolveModelWithTier", () => {
       expect(result.quotaPreference).toBe("antigravity");
     });
 
-    it("keeps antigravity for image models when cli_first is true", () => {
-      const result = resolveModelWithTier("gemini-3-pro-image", { cli_first: true });
-      expect(result.quotaPreference).toBe("antigravity");
-      expect(result.explicitQuota).toBe(true);
-    });
-
     it("defaults to antigravity when cli_first is false", () => {
       const result = resolveModelWithTier("gemini-3-flash", { cli_first: false });
       expect(result.quotaPreference).toBe("antigravity");
@@ -91,20 +85,6 @@ describe("resolveModelWithTier", () => {
   });
 
   describe("Antigravity Gemini 3 with tier suffix", () => {
-    it("antigravity-gemini-3-pro-low gets thinkingLevel from tier", () => {
-      const result = resolveModelWithTier("antigravity-gemini-3-pro-low");
-      expect(result.actualModel).toBe("gemini-3-pro-low");
-      expect(result.thinkingLevel).toBe("low");
-      expect(result.quotaPreference).toBe("antigravity");
-    });
-
-    it("antigravity-gemini-3-pro-high gets thinkingLevel from tier", () => {
-      const result = resolveModelWithTier("antigravity-gemini-3-pro-high");
-      expect(result.actualModel).toBe("gemini-3-pro-high");
-      expect(result.thinkingLevel).toBe("high");
-      expect(result.quotaPreference).toBe("antigravity");
-    });
-
     it("antigravity-gemini-3-flash-medium gets thinkingLevel from tier", () => {
       const result = resolveModelWithTier("antigravity-gemini-3-flash-medium");
       expect(result.actualModel).toBe("gemini-3-flash");
@@ -119,13 +99,53 @@ describe("resolveModelWithTier", () => {
   });
 
   describe("Claude thinking models default budget", () => {
-    it("antigravity-claude-opus-4-6-thinking gets default max budget (32768)", () => {
+    it("antigravity-claude-opus-4-6-thinking gets default high budget (32768)", () => {
       const result = resolveModelWithTier("antigravity-claude-opus-4-6-thinking");
       expect(result.actualModel).toBe("claude-opus-4-6-thinking");
       expect(result.thinkingBudget).toBe(32768);
       expect(result.isThinkingModel).toBe(true);
       expect(result.quotaPreference).toBe("antigravity");
     });
+
+    it("antigravity-claude-sonnet-4-6-thinking gets default max budget (32768)", () => {
+      const result = resolveModelWithTier("antigravity-claude-sonnet-4-6-thinking");
+      expect(result.actualModel).toBe("claude-sonnet-4-6-thinking");
+      expect(result.thinkingBudget).toBe(32768);
+      expect(result.isThinkingModel).toBe(true);
+      expect(result.quotaPreference).toBe("antigravity");
+    });
+
+    it("antigravity-claude-opus-4-6-thinking-max gets max budget (65536)", () => {
+      const result = resolveModelWithTier("antigravity-claude-opus-4-6-thinking-max");
+      expect(result.actualModel).toBe("claude-opus-4-6-thinking");
+      expect(result.thinkingBudget).toBe(65536);
+      expect(result.isThinkingModel).toBe(true);
+      expect(result.quotaPreference).toBe("antigravity");
+    });
+
+    it("antigravity-claude-sonnet-4-6-thinking-low gets low budget (8192)", () => {
+      const result = resolveModelWithTier("antigravity-claude-sonnet-4-6-thinking-low");
+      expect(result.actualModel).toBe("claude-sonnet-4-6-thinking");
+      expect(result.thinkingBudget).toBe(8192);
+      expect(result.isThinkingModel).toBe(true);
+      expect(result.quotaPreference).toBe("antigravity");
+    });
+
+    it("antigravity-claude-sonnet-4-6-thinking-medium gets medium budget (16384)", () => {
+      const result = resolveModelWithTier("antigravity-claude-sonnet-4-6-thinking-medium");
+      expect(result.actualModel).toBe("claude-sonnet-4-6-thinking");
+      expect(result.thinkingBudget).toBe(16384);
+      expect(result.isThinkingModel).toBe(true);
+      expect(result.quotaPreference).toBe("antigravity");
+    });
+
+    it("antigravity-claude-sonnet-4-6-thinking-high gets high budget (32768)", () => {
+      const result = resolveModelWithTier("antigravity-claude-sonnet-4-6-thinking-high");
+      expect(result.actualModel).toBe("claude-sonnet-4-6-thinking");
+      expect(result.thinkingBudget).toBe(32768);
+      expect(result.isThinkingModel).toBe(true);
+      expect(result.quotaPreference).toBe("antigravity");
+    });
   });
 
   describe("Claude Sonnet 4.6 (non-thinking)", () => {
@@ -155,17 +175,9 @@ describe("resolveModelWithTier", () => {
   });
 
   describe("Image models", () => {
-    it("marks antigravity-gemini-3-pro-image as explicit quota", () => {
-      const result = resolveModelWithTier("antigravity-gemini-3-pro-image");
-      expect(result.actualModel).toBe("gemini-3-pro-image");
-      expect(result.isImageModel).toBe(true);
-      expect(result.explicitQuota).toBe(true);
-      expect(result.quotaPreference).toBe("antigravity");
-    });
-
-    it("marks gemini-3-pro-image as explicit quota", () => {
-      const result = resolveModelWithTier("gemini-3-pro-image");
-      expect(result.actualModel).toBe("gemini-3-pro-image");
+    it("marks image models as explicit quota", () => {
+      const result = resolveModelWithTier("gemini-3-flash-image");
+      expect(result.actualModel).toBe("gemini-3-flash-image");
       expect(result.isImageModel).toBe(true);
       expect(result.explicitQuota).toBe(true);
       expect(result.quotaPreference).toBe("antigravity");
@@ -182,12 +194,6 @@ describe("resolveModelWithVariant", () => {
       expect(result.configSource).toBeUndefined();
     });
 
-    it("falls back to tier resolution for Gemini 3 models", () => {
-      const result = resolveModelWithVariant("gemini-3-pro-high");
-      expect(result.actualModel).toBe("gemini-3-pro");
-      expect(result.thinkingLevel).toBe("high");
-      expect(result.configSource).toBeUndefined();
-    });
   });
 
   describe("with variant config", () => {
@@ -200,16 +206,6 @@ describe("resolveModelWithVariant", () => {
       expect(result.configSource).toBe("variant");
     });
 
-    it("maps budget to thinkingLevel for Gemini 3 - low", () => {
-      const result = resolveModelWithVariant("antigravity-gemini-3-pro", {
-        thinkingBudget: 8000,
-      });
-      expect(result.actualModel).toBe("gemini-3-pro-low");
-      expect(result.thinkingLevel).toBe("low");
-      expect(result.thinkingBudget).toBeUndefined();
-      expect(result.configSource).toBe("variant");
-    });
-
     it("maps budget to thinkingLevel for Gemini 3 Flash - medium (no tier suffix)", () => {
       const result = resolveModelWithVariant("antigravity-gemini-3-flash", {
         thinkingBudget: 12000,
@@ -219,14 +215,6 @@ describe("resolveModelWithVariant", () => {
       expect(result.configSource).toBe("variant");
     });
 
-    it("maps budget to thinkingLevel for Gemini 3 - high", () => {
-      const result = resolveModelWithVariant("antigravity-gemini-3-pro", {
-        thinkingBudget: 32000,
-      });
-      expect(result.thinkingLevel).toBe("high");
-      expect(result.configSource).toBe("variant");
-    });
-
     it("uses budget directly for non-Gemini 3 models", () => {
       const result = resolveModelWithVariant("gemini-2.5-pro", {
         thinkingBudget: 20000,
diff --git a/src/plugin/transform/model-resolver.ts b/src/plugin/transform/model-resolver.ts
index dd635c3..82b7556 100644
--- a/src/plugin/transform/model-resolver.ts
+++ b/src/plugin/transform/model-resolver.ts
@@ -1,7 +1,7 @@
 /**
  * Model Resolution with Thinking Tier Support
  * 
- * Resolves model names with tier suffixes (e.g., gemini-3-pro-high, claude-opus-4-6-thinking-low)
+ * Resolves model names with tier suffixes (e.g., gemini-3.1-pro-high, claude-opus-4-6-thinking-low)
  * to their actual API model names and corresponding thinking configurations.
  */
 
@@ -16,7 +16,7 @@ export interface ModelResolverOptions {
  * Claude and Gemini 2.5 Pro use numeric budgets.
  */
 export const THINKING_TIER_BUDGETS = {
-  claude: { low: 8192, medium: 16384, high: 32768 },
+  claude: { low: 8192, medium: 16384, high: 32768, max: 65536 },
   "gemini-2.5-pro": { low: 8192, medium: 16384, high: 32768 },
   "gemini-2.5-flash": { low: 6144, medium: 12288, high: 24576 },
   default: { low: 4096, medium: 8192, high: 16384 },
@@ -33,15 +33,13 @@ export const GEMINI_3_THINKING_LEVELS = ["minimal", "low", "medium", "high"] as
  * Model aliases - maps user-friendly names to API model names.
  * 
  * Format:
- * - Gemini 3 Pro variants: gemini-3-pro-{low,medium,high}
+ * - Gemini 3 Pro variants: gemini-3.1-pro-{low,medium,high}
  * - Claude thinking variants: claude-{model}-thinking-{low,medium,high}
  * - Claude non-thinking: claude-{model} (no -thinking suffix)
  */
 export const MODEL_ALIASES: Record<string, string> = {
   // Gemini 3 variants - for Gemini CLI only (tier stripped, thinkingLevel used)
   // For Antigravity, these are bypassed and full model name is kept
-  "gemini-3-pro-low": "gemini-3-pro",
-  "gemini-3-pro-high": "gemini-3-pro",
   "gemini-3.1-pro-low": "gemini-3.1-pro",
   "gemini-3.1-pro-high": "gemini-3.1-pro",
   "gemini-3-flash-low": "gemini-3-flash",
@@ -52,6 +50,10 @@ export const MODEL_ALIASES: Record<string, string> = {
   "gemini-claude-opus-4-6-thinking-low": "claude-opus-4-6-thinking",
   "gemini-claude-opus-4-6-thinking-medium": "claude-opus-4-6-thinking",
   "gemini-claude-opus-4-6-thinking-high": "claude-opus-4-6-thinking",
+  "gemini-claude-opus-4-6-thinking-max": "claude-opus-4-6-thinking",
+  "gemini-claude-sonnet-4-6-thinking-low": "claude-sonnet-4-6-thinking",
+  "gemini-claude-sonnet-4-6-thinking-medium": "claude-sonnet-4-6-thinking",
+  "gemini-claude-sonnet-4-6-thinking-high": "claude-sonnet-4-6-thinking",
   "gemini-claude-sonnet-4-6": "claude-sonnet-4-6",
 
   // Image generation models - only gemini-3-pro-image is available via Antigravity API
@@ -59,7 +61,7 @@ export const MODEL_ALIASES: Record<string, string> = {
   // Reference: Antigravity-Manager/src-tauri/src/proxy/common/model_mapping.rs
 };
 
-const TIER_REGEX = /-(minimal|low|medium|high)$/;
+const TIER_REGEX = /-(minimal|low|medium|high|max)$/;
 const QUOTA_PREFIX_REGEX = /^antigravity-/i;
 const GEMINI_3_PRO_REGEX = /^gemini-3(?:\.\d+)?-pro/i;
 const GEMINI_3_FLASH_REGEX = /^gemini-3(?:\.\d+)?-flash/i;
@@ -149,8 +151,8 @@ function isGemini3FlashModel(model: string): boolean {
  *
  * Examples:
  * - "gemini-2.5-flash" → { quotaPreference: "antigravity" }
- * - "gemini-3-pro-preview" → { quotaPreference: "antigravity" }
- * - "antigravity-gemini-3-pro-high" → { quotaPreference: "antigravity", explicitQuota: true }
+ * - "gemini-3.1-pro-preview" → { quotaPreference: "antigravity" }
+ * - "antigravity-gemini-3.1-pro-high" → { quotaPreference: "antigravity", explicitQuota: true }
  * - "claude-opus-4-6-thinking-medium" → { quotaPreference: "antigravity" }
  *
  * @param requestedModel - The model name from the request
@@ -166,10 +168,14 @@ export function resolveModelWithTier(requestedModel: string, options: ModelResol
 
   const isImageModel = IMAGE_GENERATION_MODELS.test(modelWithoutQuota);
   const isClaudeModel = modelWithoutQuota.toLowerCase().includes("claude");
+  const isGeminiCliOnlyModel = modelWithoutQuota.toLowerCase() === "gemini-3-pro-preview";
   
   // All models default to Antigravity quota unless cli_first is enabled
   // Fallback to gemini-cli happens at the account rotation level when Antigravity is exhausted
-  const preferGeminiCli = options.cli_first === true && !isAntigravity && !isImageModel && !isClaudeModel;
+  const preferGeminiCli = (options.cli_first === true || isGeminiCliOnlyModel)
+    && !isAntigravity
+    && !isImageModel
+    && !isClaudeModel;
   const quotaPreference = preferGeminiCli ? "gemini-cli" as const : "antigravity" as const;
   const explicitQuota = isAntigravity || isImageModel;
 
@@ -177,7 +183,7 @@ export function resolveModelWithTier(requestedModel: string, options: ModelResol
   const skipAlias = isAntigravity && isGemini3;
 
   // For Antigravity Gemini 3 Pro models without explicit tier, append default tier
-  // Antigravity API: gemini-3-pro requires tier suffix (gemini-3-pro-low/high)
+  // Antigravity API: gemini-3.1-pro requires tier suffix (gemini-3.1-pro-low/high)
   //                  gemini-3-flash uses bare name + thinkingLevel param
   // Pro defaults to -low unless an explicit tier is provided
   const isGemini3Pro = isGemini3ProModel(modelWithoutQuota);
@@ -254,7 +260,7 @@ export function resolveModelWithTier(requestedModel: string, options: ModelResol
 
   const budgetFamily = getBudgetFamily(resolvedModel);
   const budgets = THINKING_TIER_BUDGETS[budgetFamily];
-  const thinkingBudget = budgets[tier];
+  const thinkingBudget = (budgets as Record<string, number>)[tier];
 
   return {
     actualModel: resolvedModel,
@@ -304,7 +310,7 @@ function budgetToGemini3Level(budget: number): "low" | "medium" | "high" {
  * 
  * Issue #103: When quota fallback occurs, model names need to be transformed:
  * - gemini-3-flash-preview (gemini-cli) → gemini-3-flash (antigravity)
- * - gemini-3-pro-preview (gemini-cli) → gemini-3-pro-low (antigravity)
+ * - gemini-3.1-pro-preview (gemini-cli) → gemini-3.1-pro-low (antigravity)
  * - gemini-3-flash (antigravity) → gemini-3-flash-preview (gemini-cli)
  */
 export function resolveModelForHeaderStyle(
@@ -325,7 +331,7 @@ export function resolveModelForHeaderStyle(
       .replace(/^antigravity-/i, "");
     
     const isGemini3Pro = isGemini3ProModel(transformedModel);
-    const hasTierSuffix = /-(low|medium|high)$/i.test(transformedModel);
+    const hasTierSuffix = /-(low|medium|high|max)$/i.test(transformedModel);
     const isImageModel = IMAGE_GENERATION_MODELS.test(transformedModel);
     
     // Don't add tier suffix to image models - they don't support thinking
@@ -340,7 +346,7 @@ export function resolveModelForHeaderStyle(
   if (headerStyle === "gemini-cli") {
     let transformedModel = requestedModel
       .replace(/^antigravity-/i, "")
-      .replace(/-(low|medium|high)$/i, "");
+      .replace(/-(low|medium|high|max)$/i, "");
 
     const hasPreviewSuffix = /-preview($|-)/i.test(transformedModel);
     if (!hasPreviewSuffix) {
@@ -390,7 +396,7 @@ export function resolveModelWithVariant(
 
     let actualModel = base.actualModel;
     if (isAntigravityGemini3Pro) {
-      const baseModel = base.actualModel.replace(/-(low|medium|high)$/, "");
+      const baseModel = base.actualModel.replace(/-(low|medium|high|max)$/, "");
       actualModel = `${baseModel}-${level}`;
     }
 
diff --git a/src/plugin/transform/types.ts b/src/plugin/transform/types.ts
index a86753b..1770e52 100644
--- a/src/plugin/transform/types.ts
+++ b/src/plugin/transform/types.ts
@@ -2,7 +2,7 @@ import type { HeaderStyle } from "../../constants";
 
 export type ModelFamily = "claude" | "gemini-flash" | "gemini-pro";
 
-export type ThinkingTier = "low" | "medium" | "high";
+export type ThinkingTier = "low" | "medium" | "high" | "max";
 
 /**
  * Context for request transformation.