fix(anthropic): align #849 max_tokens map and tests with synced model metadata

Main's OpenRouter metadata sync (e14586b) retired claude-3.x / claude-4 /
-fast model ids and added Fable 5 / Sonnet 5, so after rebasing:

- ANTHROPIC_MODEL_MAX_OUTPUT_TOKENS again mirrors ANTHROPIC_MODELS exactly
  (removed retired entries, added claude-fable-5 / claude-sonnet-5).
- Tests use current ids (claude-opus-4-8, claude-opus-4-1, claude-sonnet-4-5,
  claude-opus-4-5 for the 32K ceiling case).
- The 'ceiling below the non-streaming limit' test relied on claude-3-haiku's
  4K ceiling; no current model sits under 21K, so it now asserts every model
  in the lineup clamps on the non-streaming path instead.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Author: tombeckenham

packages/ai-anthropic/src/model-meta.ts

@@ -537,16 +537,10 @@ const ANTHROPIC_MODEL_MAX_OUTPUT_TOKENS: Record<string, number> = {
   [CLAUDE_SONNET_4_5.id]: CLAUDE_SONNET_4_5.max_output_tokens,
   [CLAUDE_HAIKU_4_5.id]: CLAUDE_HAIKU_4_5.max_output_tokens,
   [CLAUDE_OPUS_4_1.id]: CLAUDE_OPUS_4_1.max_output_tokens,
-  [CLAUDE_SONNET_4.id]: CLAUDE_SONNET_4.max_output_tokens,
-  [CLAUDE_SONNET_3_7.id]: CLAUDE_SONNET_3_7.max_output_tokens,
-  [CLAUDE_OPUS_4.id]: CLAUDE_OPUS_4.max_output_tokens,
-  [CLAUDE_HAIKU_3_5.id]: CLAUDE_HAIKU_3_5.max_output_tokens,
-  [CLAUDE_HAIKU_3.id]: CLAUDE_HAIKU_3.max_output_tokens,
-  [CLAUDE_OPUS_4_6_FAST.id]: CLAUDE_OPUS_4_6_FAST.max_output_tokens,
   [CLAUDE_OPUS_4_7.id]: CLAUDE_OPUS_4_7.max_output_tokens,
-  [CLAUDE_OPUS_4_7_FAST.id]: CLAUDE_OPUS_4_7_FAST.max_output_tokens,
   [CLAUDE_OPUS_4_8.id]: CLAUDE_OPUS_4_8.max_output_tokens,
-  [CLAUDE_OPUS_4_8_FAST.id]: CLAUDE_OPUS_4_8_FAST.max_output_tokens,
+  [CLAUDE_FABLE_5.id]: CLAUDE_FABLE_5.max_output_tokens,
+  [CLAUDE_SONNET_5.id]: CLAUDE_SONNET_5.max_output_tokens,
 }
 
 /**

packages/ai-anthropic/tests/anthropic-adapter.test.ts

@@ -464,7 +464,7 @@ describe('Anthropic adapter option mapping', () => {
     }
 
     const [payload] = mocks.betaMessagesCreate.mock.calls[0]!
-    // claude-3-7-sonnet's model-meta max_output_tokens is 64_000 — not the old
+    // claude-opus-4-1's model-meta max_output_tokens is 64_000 — not the old
     // hard-coded 1024 floor that silently truncated long responses.
     expect(payload.max_tokens).toBe(64_000)
   })
@@ -492,7 +492,7 @@ describe('Anthropic adapter option mapping', () => {
     })()
     mocks.betaMessagesCreate.mockResolvedValueOnce(truncatedStream)
 
-    const adapter = createAdapter('claude-3-7-sonnet')
+    const adapter = createAdapter('claude-opus-4-1')
 
     const logger = {
       debug: vi.fn(),
@@ -526,7 +526,7 @@ describe('Anthropic adapter option mapping', () => {
     })()
     mocks.betaMessagesCreate.mockResolvedValueOnce(truncatedStream)
 
-    const adapter = createAdapter('claude-3-7-sonnet')
+    const adapter = createAdapter('claude-opus-4-1')
 
     const logger = {
       debug: vi.fn(),
@@ -560,7 +560,7 @@ describe('Anthropic adapter option mapping', () => {
       id: 'msg_structured',
       type: 'message',
       role: 'assistant',
-      model: 'claude-3-7-sonnet',
+      model: 'claude-opus-4-1',
       content: [
         {
           type: 'tool_use',
@@ -573,7 +573,7 @@ describe('Anthropic adapter option mapping', () => {
       usage: { input_tokens: 10, output_tokens: 20 },
     })
 
-    const adapter = createAdapter('claude-3-7-sonnet')
+    const adapter = createAdapter('claude-opus-4-1')
 
     for await (const _ of chat({
       adapter,
@@ -589,7 +589,7 @@ describe('Anthropic adapter option mapping', () => {
 
     const [payload] = mocks.betaMessagesCreate.mock.calls[0]!
     expect(payload.stream).toBe(false)
-    // Clamped to the non-streaming limit — NOT claude-3-7-sonnet's full 64K
+    // Clamped to the non-streaming limit — NOT claude-opus-4-1's full 64K
     // streaming ceiling, which would make the SDK throw before the request.
     expect(payload.max_tokens).toBe(ANTHROPIC_MAX_NONSTREAMING_TOKENS)
     expect(payload.max_tokens).toBeLessThanOrEqual(21_333)

packages/ai-anthropic/tests/model-meta.test.ts

@@ -2,6 +2,7 @@ import { describe, expect, expectTypeOf, it } from 'vitest'
 import {
   ANTHROPIC_DEFAULT_MAX_OUTPUT_TOKENS,
   ANTHROPIC_MAX_NONSTREAMING_TOKENS,
+  ANTHROPIC_MODELS,
   getAnthropicDefaultMaxTokens,
 } from '../src/model-meta'
 import type {
@@ -333,11 +334,11 @@ describe('Anthropic Model Input Modality Type Assertions', () => {
 
 describe('getAnthropicDefaultMaxTokens (#849)', () => {
   it("returns the model's max_output_tokens for known models", () => {
-    expect(getAnthropicDefaultMaxTokens('claude-opus-4.8')).toBe(128_000)
+    expect(getAnthropicDefaultMaxTokens('claude-opus-4-8')).toBe(128_000)
     expect(getAnthropicDefaultMaxTokens('claude-opus-4-6')).toBe(128_000)
     expect(getAnthropicDefaultMaxTokens('claude-sonnet-4-6')).toBe(64_000)
-    expect(getAnthropicDefaultMaxTokens('claude-3-7-sonnet')).toBe(64_000)
-    expect(getAnthropicDefaultMaxTokens('claude-3-haiku')).toBe(4_000)
+    expect(getAnthropicDefaultMaxTokens('claude-sonnet-4-5')).toBe(64_000)
+    expect(getAnthropicDefaultMaxTokens('claude-opus-4-5')).toBe(32_000)
   })
 
   it('falls back to the safe constant for unknown models', () => {
@@ -348,7 +349,7 @@ describe('getAnthropicDefaultMaxTokens (#849)', () => {
   })
 
   it('never returns the old hard-coded 1024 floor for a known model', () => {
-    expect(getAnthropicDefaultMaxTokens('claude-opus-4.8')).toBeGreaterThan(
+    expect(getAnthropicDefaultMaxTokens('claude-opus-4-8')).toBeGreaterThan(
       1024,
     )
   })
@@ -361,7 +362,7 @@ describe('getAnthropicDefaultMaxTokens (#849)', () => {
 
     // Opus 128K and Sonnet 64K both exceed the non-streaming limit → clamped.
     expect(
-      getAnthropicDefaultMaxTokens('claude-opus-4.8', { stream: false }),
+      getAnthropicDefaultMaxTokens('claude-opus-4-8', { stream: false }),
     ).toBe(ANTHROPIC_MAX_NONSTREAMING_TOKENS)
     expect(
       getAnthropicDefaultMaxTokens('claude-sonnet-4-6', { stream: false }),
@@ -374,19 +375,22 @@ describe('getAnthropicDefaultMaxTokens (#849)', () => {
     ).toBe(ANTHROPIC_MAX_NONSTREAMING_TOKENS)
   })
 
-  it('does not clamp a model whose ceiling is already below the non-streaming limit (#849)', () => {
-    // claude-3-haiku's 4K ceiling is under the non-streaming limit, so the
-    // non-streaming path returns the real ceiling, not the (larger) cap.
-    expect(
-      getAnthropicDefaultMaxTokens('claude-3-haiku', { stream: false }),
-    ).toBe(4_000)
+  it('clamps every current model on the non-streaming path (#849)', () => {
+    // Every model in today's lineup has a ceiling (32K–128K) above the
+    // non-streaming limit, so all of them clamp. (The clamp is a Math.min, so
+    // a future sub-21K model would keep its real ceiling instead.)
+    for (const model of ANTHROPIC_MODELS) {
+      expect(getAnthropicDefaultMaxTokens(model, { stream: false })).toBe(
+        ANTHROPIC_MAX_NONSTREAMING_TOKENS,
+      )
+    }
   })
 
   it('keeps the full ceiling for streaming requests (default) (#849)', () => {
     expect(
-      getAnthropicDefaultMaxTokens('claude-opus-4.8', { stream: true }),
+      getAnthropicDefaultMaxTokens('claude-opus-4-8', { stream: true }),
     ).toBe(128_000)
     // Omitting the option defaults to streaming.
-    expect(getAnthropicDefaultMaxTokens('claude-opus-4.8')).toBe(128_000)
+    expect(getAnthropicDefaultMaxTokens('claude-opus-4-8')).toBe(128_000)
   })
 })