codeep 2.8.0 → 2.9.0

package/README.md

@@ -50,7 +50,7 @@ custom slash commands, lifecycle hooks, checkpoints, `/cost`,
 ### Multi-Provider Support
 - **Z.AI (ZhipuAI)** — GLM models (Coding Plan & pay-per-use API, international & China)
 - **OpenAI** — GPT models (flagship, Mini, Nano)
-- **Anthropic** — Claude models (Opus, Sonnet, Haiku)
+- **Anthropic** — Claude models (Fable, Opus, Sonnet, Haiku)
 - **DeepSeek** — DeepSeek models (Pro, Flash)
 - **Google AI** — Gemini models (Pro, Flash)
 - **MiniMax** — MiniMax models (Coding Plan & pay-per-use API, international & China)

package/dist/api/index.js

@@ -2,7 +2,7 @@ import * as http from 'node:http';
 import * as https from 'node:https';
 import { config, getApiKey, resolveBaseUrl } from '../config/index.js';
 import { withRetry, isNetworkError } from '../utils/retry.js';
-import { getProvider, getProviderBaseUrl, getProviderAuthHeader, usesMaxCompletionTokens, requiresDefaultTemperature } from '../config/providers.js';
+import { getProvider, getProviderBaseUrl, getProviderAuthHeader, usesMaxCompletionTokens, requiresDefaultTemperature, modelRejectsSamplingParams } from '../config/providers.js';
 import { logApiRequest, logApiResponse } from '../utils/logger.js';
 import { loadProjectIntelligence, generateContextFromIntelligence } from '../utils/projectIntelligence.js';
 import { loadProjectRules } from '../utils/agent.js';
@@ -655,7 +655,9 @@ async function chatAnthropic(message, history, model, apiKey, onChunk, abortSign
                 model,
                 messages,
                 max_tokens: maxTokens,
-                temperature,
+                // Fable 5 / Opus 4.7+ reject temperature with a 400 — omit it there
+                // (omission means API default on every Claude model).
+                ...(modelRejectsSamplingParams(model) ? {} : { temperature }),
                 stream,
                 ...cachedSystem,
             }),

package/dist/config/providers.d.ts

@@ -68,6 +68,7 @@ export declare function usesMaxCompletionTokens(providerId: string): boolean;
  * (e.g. OpenAI GPT-5+ only accepts the default of 1).
  */
 export declare function requiresDefaultTemperature(providerId: string): boolean;
+export declare function modelRejectsSamplingParams(model: string): boolean;
 /**
  * Returns the effective max output tokens for a provider, capped by the provider's limit.
  * Falls back to the requested value if no provider limit is set.

package/dist/config/providers.js

@@ -243,9 +243,8 @@ export const PROVIDERS = {
             },
         },
         models: [
-            { id: 'claude-opus-4-8', name: 'Claude Opus 4.8', description: 'Most capable Claude model' },
-            { id: 'claude-opus-4-7', name: 'Claude Opus 4.7', description: 'Previous generation Opus' },
-            { id: 'claude-opus-4-6', name: 'Claude Opus 4.6', description: 'Older generation Opus' },
+            { id: 'claude-fable-5', name: 'Claude Fable 5', description: 'Most powerful — new tier above Opus' },
+            { id: 'claude-opus-4-8', name: 'Claude Opus 4.8', description: 'Most capable Opus model' },
             { id: 'claude-sonnet-4-6', name: 'Claude Sonnet', description: 'Best balance of speed and intelligence' },
             { id: 'claude-haiku-4-5-20251001', name: 'Claude Haiku', description: 'Fastest and most affordable' },
         ],
@@ -448,6 +447,17 @@ export function usesMaxCompletionTokens(providerId) {
 export function requiresDefaultTemperature(providerId) {
     return PROVIDERS[providerId]?.requiresDefaultTemperature ?? false;
 }
+/**
+ * Models that reject sampling parameters (temperature/top_p/top_k) with a 400.
+ * Anthropic removed them on Fable 5 and Opus 4.7+; older Claude models still
+ * accept them, so this must be a MODEL-level check, not a provider-level one
+ * (requiresDefaultTemperature can't express it). Omitting the field is always
+ * safe — the API treats omission as default.
+ */
+const SAMPLING_PARAMS_REJECTED = ['claude-fable-5', 'claude-opus-4-8', 'claude-opus-4-7'];
+export function modelRejectsSamplingParams(model) {
+    return SAMPLING_PARAMS_REJECTED.some(id => model === id || model.startsWith(`${id}-`));
+}
 /**
  * Returns the effective max output tokens for a provider, capped by the provider's limit.
  * Falls back to the requested value if no provider limit is set.

package/dist/utils/agentChat.js

@@ -17,7 +17,7 @@ import { createHash } from 'crypto';
 import { config, getApiKey, resolveBaseUrl } from '../config/index.js';
 import { loadProjectIntelligence, generateContextFromIntelligence } from './projectIntelligence.js';
 import { syncProgress, generateProjectId } from './codeepCloud.js';
-import { getProviderAuthHeader, supportsNativeTools, getEffectiveMaxTokens, usesMaxCompletionTokens, requiresDefaultTemperature, isNoApiKeyProvider } from '../config/providers.js';
+import { getProviderAuthHeader, supportsNativeTools, getEffectiveMaxTokens, usesMaxCompletionTokens, requiresDefaultTemperature, modelRejectsSamplingParams, isNoApiKeyProvider } from '../config/providers.js';
 import { recordTokenUsage, extractOpenAIUsage, extractAnthropicUsage } from './tokenTracker.js';
 import { parseOpenAIToolCalls, parseAnthropicToolCalls, parseToolCalls } from './toolParsing.js';
 import { formatToolDefinitions, getOpenAITools, getAnthropicTools } from './tools.js';
@@ -357,7 +357,9 @@ additionalTools) {
         let endpoint;
         let body;
         const useStreaming = Boolean(onChunk);
-        const tempParam = requiresDefaultTemperature(providerId) ? {} : { temperature: config.get('temperature') };
+        // Provider-level guard (OpenAI GPT-5+) OR model-level guard — Anthropic's
+        // Fable 5 / Opus 4.7+ reject temperature with a 400; omission is safe.
+        const tempParam = (requiresDefaultTemperature(providerId) || modelRejectsSamplingParams(model)) ? {} : { temperature: config.get('temperature') };
         if (protocol === 'openai') {
             const maxTok = getEffectiveMaxTokens(providerId, Math.max(config.get('maxTokens'), 16384));
             const tokParam = usesMaxCompletionTokens(providerId) ? { max_completion_tokens: maxTok } : { max_tokens: maxTok };
@@ -549,7 +551,9 @@ export async function agentChatFallback(messages, systemPrompt, onChunk, abortSi
     try {
         let endpoint;
         let body;
-        const tempParam = requiresDefaultTemperature(providerId) ? {} : { temperature: config.get('temperature') };
+        // Provider-level guard (OpenAI GPT-5+) OR model-level guard — Anthropic's
+        // Fable 5 / Opus 4.7+ reject temperature with a 400; omission is safe.
+        const tempParam = (requiresDefaultTemperature(providerId) || modelRejectsSamplingParams(model)) ? {} : { temperature: config.get('temperature') };
         if (protocol === 'openai') {
             const maxTok = getEffectiveMaxTokens(providerId, Math.max(config.get('maxTokens'), 16384));
             const tokParam = usesMaxCompletionTokens(providerId) ? { max_completion_tokens: maxTok } : { max_tokens: maxTok };

package/dist/utils/tokenTracker.js

@@ -16,9 +16,8 @@ const MODEL_CONTEXT_WINDOWS = {
     'gpt-5.4-mini': 400_000,
     'gpt-5.4-nano': 400_000,
     // Anthropic
+    'claude-fable-5': 1_000_000,
     'claude-opus-4-8': 1_000_000,
-    'claude-opus-4-7': 1_000_000,
-    'claude-opus-4-6': 1_000_000,
     'claude-sonnet-4-6': 1_000_000,
     'claude-haiku-4-5-20251001': 200_000,
     // DeepSeek
@@ -52,9 +51,8 @@ const MODEL_PRICING = {
     'gpt-5.4-mini': { inputPer1M: 0.75, outputPer1M: 4.50 },
     'gpt-5.4-nano': { inputPer1M: 0.20, outputPer1M: 1.25 },
     // Anthropic
+    'claude-fable-5': { inputPer1M: 10.00, outputPer1M: 50.00 },
     'claude-opus-4-8': { inputPer1M: 5.00, outputPer1M: 25.00 },
-    'claude-opus-4-7': { inputPer1M: 5.00, outputPer1M: 25.00 },
-    'claude-opus-4-6': { inputPer1M: 5.00, outputPer1M: 25.00 },
     'claude-sonnet-4-6': { inputPer1M: 3.00, outputPer1M: 15.00 },
     'claude-haiku-4-5-20251001': { inputPer1M: 1.00, outputPer1M: 5.00 },
     // DeepSeek (cache-miss input pricing)

package/dist/version.d.ts

@@ -1 +1 @@
-export declare const VERSION = "2.8.0";
+export declare const VERSION = "2.9.0";

package/dist/version.js

@@ -1,4 +1,4 @@
 // AUTO-GENERATED by scripts/gen-version.js — do not edit by hand.
 // Baked from package.json at build time so the bun-compiled binary reports
 // the right version (it has no package.json on disk to read at runtime).
-export const VERSION = '2.8.0';
+export const VERSION = '2.9.0';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "codeep",
-  "version": "2.8.0",
+  "version": "2.9.0",
   "description": "AI-powered coding assistant built for the terminal. Multiple LLM providers, project-aware context, and a seamless development workflow.",
   "type": "module",
   "main": "dist/index.js",