# mxchat-basic/3.2.24/includes/class-mxchat-model-catalog.php MxChat – AI Chatbot & Content Generation for WordPress, version 3.2.24. 947 lines. - Page: https://pluginprobe.com/plugins/mxchat-basic/3.2.24/code/includes/class-mxchat-model-catalog.php - Raw: https://pluginprobe.com/plugins/mxchat-basic/3.2.24/raw/includes/class-mxchat-model-catalog.php - Modified: 2026-09-29T10:44:16+00:00 Line numbers below start at 1. Link to a line or a range by appending a fragment to the page URL, for example `https://pluginprobe.com/plugins/mxchat-basic/3.2.24/code/includes/class-mxchat-model-catalog.php#L10-L20`. ```php [ * 'label' => Display name shown in pickers, * 'key_option' => sub-key under get_option('mxchat_options'), * 'requires_key_to_load_models' => bool — OpenRouter is true (key needed * before model list can be fetched). Surfaces consult * this to disable / gate the provider in pickers that * can't run the inline key-then-fetch flow. * 'models' => [ model_id => [ 'label' => …, 'description' => … ] ] * ] * ] * * @package MxChat * @since 3.2.7 */ if (!defined('ABSPATH')) { exit; } class MxChat_Model_Catalog { /** * Chat models grouped by provider. Order within a provider matches the * order shown in the Settings modal picker — keep the recommended / * flagship model at top. */ public static function chat_models() { return array( 'openrouter' => array( 'label' => __('OpenRouter', 'mxchat'), 'key_option' => 'openrouter_api_key', 'requires_key_to_load_models' => true, 'disable_in_onboarding' => true, 'models' => array( 'openrouter' => array( 'label' => 'OpenRouter', 'description' => __('Access 100+ models from multiple providers (add API key to browse)', 'mxchat'), ), ), ), 'gemini' => array( 'label' => __('Google Gemini', 'mxchat'), 'key_option' => 'gemini_api_key', 'requires_key_to_load_models' => false, 'models' => array( 'gemini-3.8-flash' => array('label' => 'Gemini 3.8 Flash', 'description' => __('Stable — newest Flash generation, recommended default', 'mxchat')), 'gemini-3.7-flash' => array('label' => 'Gemini 3.7 Flash', 'description' => __('Stable — efficiency-first Flash', 'mxchat')), 'gemini-3.6-flash' => array('label' => 'Gemini 3.6 Flash', 'description' => __('Stable — previous Flash generation', 'mxchat')), 'gemini-3.5-flash' => array('label' => 'Gemini 3.5 Flash', 'description' => __('Stable — older Flash, priced above 3.8 Flash on every token', 'mxchat')), 'gemini-3.1-pro-preview' => array('label' => 'Gemini 3.1 Pro', 'description' => __('Preview — most intelligent, multimodal & agentic', 'mxchat')), 'gemini-3-flash-preview' => array('label' => 'Gemini 3 Flash', 'description' => __('Preview — balanced speed and scale', 'mxchat')), 'gemini-3.5-flash-lite' => array('label' => 'Gemini 3.5 Flash-Lite', 'description' => __('Stable — cheapest Gemini, no thinking', 'mxchat')), 'gemini-3.1-flash-lite' => array('label' => 'Gemini 3.1 Flash-Lite', 'description' => __('Stable — cost-efficient, high throughput', 'mxchat')), 'gemini-3.1-flash-lite-preview' => array('label' => 'Gemini 3.1 Flash-Lite (Preview)', 'description' => __('Preview — latest cost-efficient model', 'mxchat')), 'gemini-2.5-pro' => array('label' => 'Gemini 2.5 Pro', 'description' => __('Stable — advanced thinking, code, math & long context', 'mxchat')), 'gemini-2.5-flash' => array('label' => 'Gemini 2.5 Flash', 'description' => __('Stable — strong price-performance with thinking', 'mxchat')), 'gemini-2.5-flash-lite' => array('label' => 'Gemini 2.5 Flash-Lite', 'description' => __('Stable — ultra fast and low cost', 'mxchat')), ), ), 'openai' => array( 'label' => __('OpenAI', 'mxchat'), 'key_option' => 'api_key', 'requires_key_to_load_models' => false, 'models' => array( // OpenAI retired gpt-5.1-chat-latest and gpt-5.3-chat-latest on // 2026-08-10 (replacement gpt-5.6-sol per the deprecations page); // stored settings migrate via mxchat_migrate_deprecated_models() // and the integrator carries a read-time rescue (plan e46b8f). // // 'deprecated_on' (Y-m-d) + 'replacement' are optional per-model // fields: pickers label a future date as "retires " and hide // the entry once the date passes (a saved value keeps rendering via // settings_dropdown_groups($current_model)). // GPT-6 (announced 2026-09-22; plan 45c22d). Probed live on // 2026-09-27: all three need max_completion_tokens, reject any // temperature but 1 and reject reasoning_effort 'minimal'; // Astra also rejects 'none'. Every "modern OpenAI" gate goes // through is_modern_openai() so these ids are not sent legacy // parameters (which 400 on the first message). 'gpt-6-astra' => array('label' => 'GPT-6 Astra', 'description' => __('Recommended — newest OpenAI flagship; strongest reasoning, coding and agentic work', 'mxchat')), 'gpt-6-sol' => array('label' => 'GPT-6 Sol', 'description' => __('Strong reasoning on demanding tasks at mid-tier cost', 'mxchat')), 'gpt-6-luna' => array('label' => 'GPT-6 Luna', 'description' => __('Fastest and cheapest GPT-6 for high-volume, repeatable work', 'mxchat')), 'gpt-5.6-sol' => array('label' => 'GPT-5.6 Sol', 'description' => __('Previous OpenAI flagship for reasoning, coding and chat', 'mxchat')), 'gpt-5.6-terra' => array('label' => 'GPT-5.6 Terra', 'description' => __('Balanced GPT-5.6 — strong capability at lower cost', 'mxchat')), 'gpt-5.6-luna' => array('label' => 'GPT-5.6 Luna', 'description' => __('Fastest and cheapest GPT-5.6 for lightweight tasks', 'mxchat')), 'gpt-5.5' => array('label' => 'GPT-5.5', 'description' => __('Previous flagship — powerful reasoning and coding model', 'mxchat')), 'gpt-5.4' => array('label' => 'GPT-5.4', 'description' => __('Flagship reasoning and coding model', 'mxchat')), 'gpt-5.4-mini' => array('label' => 'GPT-5.4 Mini', 'description' => __('Fast and affordable with 400K context', 'mxchat')), 'gpt-5.4-nano' => array('label' => 'GPT-5.4 Nano', 'description' => __('Fastest and cheapest for lightweight tasks', 'mxchat')), 'gpt-5.2' => array('label' => 'GPT-5.2', 'description' => __('Best general-purpose & agentic model with fast responses', 'mxchat')), 'gpt-5.1-2025-11-13' => array('label' => 'GPT-5.1', 'description' => __('Flagship for coding & agentic tasks with low reasoning (400K context)', 'mxchat')), 'gpt-5' => array('label' => 'GPT-5', 'description' => __('Flagship for coding, reasoning, and agentic tasks across domains', 'mxchat'), 'deprecated_on' => '2026-12-11', 'replacement' => 'gpt-5.6-sol'), 'gpt-5-mini' => array('label' => 'GPT-5 Mini', 'description' => __('Fast and lightweight', 'mxchat'), 'deprecated_on' => '2026-12-11', 'replacement' => 'gpt-5.6-terra'), 'gpt-5-nano' => array('label' => 'GPT-5 Nano', 'description' => __('Fastest and cheapest; ideal for summarization and classification', 'mxchat'), 'deprecated_on' => '2026-12-11', 'replacement' => 'gpt-5.6-luna'), ), ), 'claude' => array( 'label' => __('Anthropic Claude', 'mxchat'), 'key_option' => 'claude_api_key', 'requires_key_to_load_models' => false, 'models' => array( 'claude-fable-5' => array('label' => 'Claude Fable 5', 'description' => __('Latest Flagship — newest and most capable Anthropic model', 'mxchat')), 'claude-opus-5' => array('label' => 'Claude Opus 5', 'description' => __('Latest Opus — best for complex agentic and coding work', 'mxchat')), 'claude-opus-4-8' => array('label' => 'Claude Opus 4.8', 'description' => __('Previous Opus generation', 'mxchat')), 'claude-opus-4-7' => array('label' => 'Claude Opus 4.7', 'description' => __('Previous Anthropic flagship model', 'mxchat')), 'claude-opus-4-6' => array('label' => 'Claude Opus 4.6', 'description' => __('Most capable Claude model - recommended', 'mxchat')), 'claude-sonnet-5' => array('label' => 'Claude Sonnet 5', 'description' => __('Newest Sonnet - excellent balance of speed and capability', 'mxchat')), 'claude-sonnet-4-6' => array('label' => 'Claude Sonnet 4.6', 'description' => __('Previous Sonnet - excellent balance of speed and capability', 'mxchat')), 'claude-opus-4-5' => array('label' => 'Claude Opus 4.5', 'description' => __('Highly capable for complex tasks', 'mxchat')), 'claude-sonnet-4-5-20250929' => array('label' => 'Claude Sonnet 4.5', 'description' => __('Best for complex agents and coding', 'mxchat')), 'claude-haiku-4-5-20251001' => array('label' => 'Claude Haiku 4.5', 'description' => __('Fastest and most intelligent Haiku', 'mxchat')), ), ), 'xai' => array( 'label' => __('xAI Grok', 'mxchat'), 'key_option' => 'xai_api_key', 'requires_key_to_load_models' => false, // Refreshed 2026-08-14 (plan-mxchat-20260814-e1aa4a) against TWO // primary sources that agree: a live pull of xAI's /v1/models with // a real key, and docs.x.ai/docs/models. Context windows below are // that listing's own context_length; image-input support for every // current-generation id was confirmed by a 32x32 probe returning a // correct answer (all HTTP 200), not inferred from docs. // // The legacy block below is INTENTIONALLY still here. xAI has // published no retirement dates for these ids and they still answer // HTTP 200 (grok-4-0709 re-probed 2026-08-14), so 'deprecated_on' // would be an invented date and deleting them would strand every // install saved on one. Current generation simply sorts above them. 'models' => array( 'grok-4.6' => array('label' => 'Grok 4.6', 'description' => __('Newest xAI flagship — 500K context, accepts image input', 'mxchat')), 'grok-4.5' => array('label' => 'Grok 4.5', 'description' => __('Previous flagship — 500K context, accepts image input', 'mxchat')), 'grok-4.3' => array('label' => 'Grok 4.3', 'description' => __('Largest context in the Grok lineup — 1M tokens, accepts image input', 'mxchat')), 'grok-4.20-0309-reasoning' => array('label' => 'Grok 4.20 (Reasoning)', 'description' => __('1M context with reasoning — dated release, pinned behaviour', 'mxchat')), 'grok-4.20-0309-non-reasoning'=> array('label' => 'Grok 4.20 (Non-Reasoning)', 'description' => __('1M context without reasoning — faster replies at the same price', 'mxchat')), // --- Previous generations. Still served by xAI, no announced // retirement date, kept so saved selections keep working. --- 'grok-4-1-fast-reasoning' => array('label' => 'Grok 4.1 Fast (Reasoning)', 'description' => __('Previous generation — no longer listed by xAI, still served', 'mxchat')), 'grok-4-1-fast-non-reasoning' => array('label' => 'Grok 4.1 Fast (Non-Reasoning)', 'description' => __('Previous generation — no longer listed by xAI, still served', 'mxchat')), 'grok-4-0709' => array('label' => 'Grok 4', 'description' => __('Previous generation — no longer listed by xAI, still served', 'mxchat')), 'grok-3-beta' => array('label' => 'Grok-3', 'description' => __('Older generation — no longer listed by xAI, still served', 'mxchat')), 'grok-3-fast-beta' => array('label' => 'Grok-3 Fast', 'description' => __('Older generation — no longer listed by xAI, still served', 'mxchat')), 'grok-3-mini-beta' => array('label' => 'Grok-3 Mini', 'description' => __('Older generation — no longer listed by xAI, still served', 'mxchat')), 'grok-3-mini-fast-beta' => array('label' => 'Grok-3 Mini Fast', 'description' => __('Older generation — no longer listed by xAI, still served', 'mxchat')), ), ), 'deepseek' => array( 'label' => __('DeepSeek', 'mxchat'), 'key_option' => 'deepseek_api_key', 'requires_key_to_load_models' => false, 'models' => array( // DeepSeek retired deepseek-chat / deepseek-reasoner on 2026-07-24; // stored settings migrate via mxchat_migrate_deprecated_models(). 'deepseek-v4-flash' => array('label' => 'DeepSeek V4 Flash', 'description' => __('Fast and cost-effective', 'mxchat')), 'deepseek-v4-pro' => array('label' => 'DeepSeek V4 Pro', 'description' => __('Most capable DeepSeek model', 'mxchat')), ), ), 'custom' => array( 'label' => __('Custom (OpenAI-compatible)', 'mxchat'), 'key_option' => 'custom_provider_api_key', 'requires_key_to_load_models' => false, 'disable_in_onboarding' => true, 'models' => array( 'custom-provider' => array( 'label' => 'Custom Provider', 'description' => __('OpenAI-compatible local LLM (Ollama, LM Studio, vLLM, llama.cpp, Azure OpenAI) — configure Base URL, key, and model in API Keys tab', 'mxchat'), ), ), ), ); } /** * Embedding models grouped by provider. Order within a provider matches * the order shown in the Settings dropdown. */ public static function embedding_models() { return array( 'openai' => array( 'label' => __('OpenAI', 'mxchat'), 'key_option' => 'api_key', 'models' => array( 'text-embedding-ada-002' => array('label' => 'Ada 2', 'description' => __('1536 dim, recommended', 'mxchat')), 'text-embedding-3-small' => array('label' => 'TE3 Small', 'description' => __('1536 dim, efficient', 'mxchat')), 'text-embedding-3-large' => array('label' => 'TE3 Large', 'description' => __('3072 dim, powerful', 'mxchat')), ), ), 'voyage' => array( 'label' => __('Voyage AI', 'mxchat'), 'key_option' => 'voyage_api_key', 'models' => array( 'voyage-3-large' => array('label' => 'Voyage-3 Large', 'description' => __('2048 dim, most capable', 'mxchat')), ), ), 'gemini' => array( 'label' => __('Google Gemini', 'mxchat'), 'key_option' => 'gemini_api_key', 'models' => array( 'gemini-embedding-001' => array('label' => 'Gemini Embedding', 'description' => __('1536 dim', 'mxchat')), ), ), ); } /** * Canonical retired-model map: every chat model id a provider has retired * (or hard-deprecated) that stored settings may still carry, mapped to its * current replacement (plan-mxchat-20260822-202df5). * * THE ONLY PLACE THIS MAPPING LIVES. mxchat_migrate_deprecated_models() * consumes it for the main chat + content models, and add-ons that store * their own model ids (mxchat-multi-bot's per-bot overrides) consume it for * theirs — a second hardcoded list in any consumer is how a stranded-model * bug recurs, so route every new retirement through this map. * * Claude ids map BY TIER so a rescued site lands on the current generation, * not an older intermediate (plan dc91bd). Never map onto a model that is * itself on a deprecation list (the e46b8f rule). * * Entry shape: retired_id => array( * 'to' => replacement model id (must be current, never itself retired), * 'label' => replacement display name for user-facing notices, * 'reason' => notice clause explaining the retirement; deliberately NOT * translated — migration messages are stored in an option and * rendered later, matching the existing migration notices. * ) */ public static function retired_model_map() { return array( // Anthropic Claude — Opus tier. 'claude-3-opus-20240229' => array('to' => 'claude-opus-5', 'label' => 'Claude Opus 5', 'reason' => 'because Anthropic retired the older Claude model'), // Retired Jan 5, 2026 'claude-opus-4-20250514' => array('to' => 'claude-opus-5', 'label' => 'Claude Opus 5', 'reason' => 'because Anthropic retired the older Claude model'), // Deprecated 'claude-opus-4-1-20250805' => array('to' => 'claude-opus-5', 'label' => 'Claude Opus 5', 'reason' => 'because Anthropic retired the older Claude model'), // Retired Aug 5, 2026 (first-party API) // Anthropic Claude — Sonnet tier. 'claude-3-5-sonnet-20240620' => array('to' => 'claude-sonnet-5', 'label' => 'Claude Sonnet 5', 'reason' => 'because Anthropic retired the older Claude model'), // Retired Oct 28, 2025 'claude-3-5-sonnet-20241022' => array('to' => 'claude-sonnet-5', 'label' => 'Claude Sonnet 5', 'reason' => 'because Anthropic retired the older Claude model'), // Retired Oct 28, 2025 'claude-3-7-sonnet-20250219' => array('to' => 'claude-sonnet-5', 'label' => 'Claude Sonnet 5', 'reason' => 'because Anthropic retired the older Claude model'), // Retired Feb 19, 2026 'claude-3-sonnet-20240229' => array('to' => 'claude-sonnet-5', 'label' => 'Claude Sonnet 5', 'reason' => 'because Anthropic retired the older Claude model'), // Legacy 'claude-sonnet-4-20250514' => array('to' => 'claude-sonnet-5', 'label' => 'Claude Sonnet 5', 'reason' => 'because Anthropic retired the older Claude model'), // Deprecated // Anthropic Claude — Haiku tier. 'claude-3-haiku-20240307' => array('to' => 'claude-haiku-4-5-20251001', 'label' => 'Claude Haiku 4.5', 'reason' => 'because Anthropic retired the older Claude model'), // Legacy 'claude-3-5-haiku-20241022' => array('to' => 'claude-haiku-4-5-20251001', 'label' => 'Claude Haiku 4.5', 'reason' => 'because Anthropic retired the older Claude model'), // Deprecated // OpenAI — GPT-4 series + GPT-3.5 Turbo (deprecated 2026-02-17). 'gpt-4o' => array('to' => 'gpt-5.6-sol', 'label' => 'GPT-5.6 Sol', 'reason' => 'due to OpenAI deprecating older models'), 'gpt-4.1-2025-04-14' => array('to' => 'gpt-5.6-sol', 'label' => 'GPT-5.6 Sol', 'reason' => 'due to OpenAI deprecating older models'), 'gpt-4-turbo' => array('to' => 'gpt-5.6-sol', 'label' => 'GPT-5.6 Sol', 'reason' => 'due to OpenAI deprecating older models'), 'gpt-4' => array('to' => 'gpt-5.6-sol', 'label' => 'GPT-5.6 Sol', 'reason' => 'due to OpenAI deprecating older models'), 'gpt-3.5-turbo' => array('to' => 'gpt-5.6-sol', 'label' => 'GPT-5.6 Sol', 'reason' => 'due to OpenAI deprecating older models'), // OpenAI — the gpt-5.x-chat-latest aliases retired 2026-08-10 (e46b8f). 'gpt-5.1-chat-latest' => array('to' => 'gpt-5.6-sol', 'label' => 'GPT-5.6 Sol', 'reason' => 'because OpenAI retires the GPT-5.x Chat Latest models on August 10, 2026'), 'gpt-5.3-chat-latest' => array('to' => 'gpt-5.6-sol', 'label' => 'GPT-5.6 Sol', 'reason' => 'because OpenAI retires the GPT-5.x Chat Latest models on August 10, 2026'), // OpenAI — mini tier. 'gpt-4o-mini' => array('to' => 'gpt-5-mini', 'label' => 'GPT-5 Mini', 'reason' => 'due to OpenAI deprecating GPT-4 series models'), 'gpt-4.1-mini' => array('to' => 'gpt-5-mini', 'label' => 'GPT-5 Mini', 'reason' => 'due to OpenAI deprecating GPT-4 series models'), // DeepSeek — removed at the vendor 2026-07-24 (hard cutoff, every request 400s). 'deepseek-chat' => array('to' => 'deepseek-v4-flash', 'label' => 'DeepSeek V4 Flash', 'reason' => 'because DeepSeek retired its older API models on July 24, 2026'), 'deepseek-reasoner' => array('to' => 'deepseek-v4-flash', 'label' => 'DeepSeek V4 Flash', 'reason' => 'because DeepSeek retired its older API models on July 24, 2026'), ); } /** * Returns the display label for the given chat-or-embedding provider slug. * Chat providers checked first; embedding-only providers fall through. */ public static function provider_label($slug) { $chat = self::chat_models(); if (isset($chat[$slug])) return $chat[$slug]['label']; $emb = self::embedding_models(); if (isset($emb[$slug])) return $emb[$slug]['label']; return $slug; } /** * Returns the mxchat_options sub-key that holds the API key for the given * provider slug. Chat providers checked first; embedding-only providers * (voyage) fall through. */ public static function key_option_for_provider($slug) { $chat = self::chat_models(); if (isset($chat[$slug])) return $chat[$slug]['key_option']; $emb = self::embedding_models(); if (isset($emb[$slug])) return $emb[$slug]['key_option']; return ''; } /** * The provider slug that owns a chat model id, or '' when the catalog * doesn't list it (plan b65e8d). * * NOTE for callers doing retirement work: a model the provider has already * retired is frequently ABSENT from this catalog while still saved on an * install — that is precisely the state the liveness check exists to catch. * Treat '' as "catalog doesn't know", not "not a real model", and fall back * to a family-prefix guess if you need one. */ public static function provider_for_chat_model($model_id) { $model_id = (string) $model_id; foreach (self::chat_models() as $provider_slug => $provider) { if (isset($provider['models'][$model_id])) { return $provider_slug; } } return ''; } /** * URL + auth headers for a provider's public model-listing endpoint * (plan b65e8d — the daily liveness check). * * Only providers that actually publish a listing are mapped. OpenRouter is * deliberately excluded: its 'openrouter' catalog entry is a meta-selector * that resolves to a per-account sub-model at request time, so "is the * configured id listed" is not a meaningful question. DeepSeek and the * custom OpenAI-compatible provider are excluded too — a self-hosted or * proxied endpoint's /models is not authoritative about retirement. * * @param string $provider Provider slug. * @param string $api_key The install's key for that provider. * @return array|null array('url' => string, 'headers' => array) or null. */ public static function models_listing_request($provider, $api_key) { $api_key = (string) $api_key; if ($api_key === '') { return null; } switch ($provider) { case 'openai': return array( 'url' => 'https://api.openai.com/v1/models', 'headers' => array('Authorization' => 'Bearer ' . $api_key), ); case 'claude': return array( 'url' => 'https://api.anthropic.com/v1/models?limit=1000', 'headers' => array( 'x-api-key' => $api_key, 'anthropic-version' => '2023-06-01', ), ); case 'gemini': return array( 'url' => 'https://generativelanguage.googleapis.com/v1beta/models?pageSize=1000', 'headers' => array('x-goog-api-key' => $api_key), ); case 'xai': return array( 'url' => 'https://api.x.ai/v1/models', 'headers' => array('Authorization' => 'Bearer ' . $api_key), ); } return null; } /** * A minimal 1-token chat request used ONLY to confirm whether a model id * still exists (plan b65e8d). Never used for content — the response is * thrown away; only the HTTP status and error code are read. * * WHY THIS EXISTS: a provider's model LISTING is not authoritative about * retirement, in either direction (all probed live 2026-08-13): * - OpenAI KEEPS retired ids listed — gpt-5.3-chat-latest is in * /v1/models and 404s "model_deprecated" on use. * - Anthropic DROPS working aliases — claude-opus-4-5 is absent from * /v1/models and answers HTTP 200 perfectly. * - xAI's listing is partial — grok-4-0709 is absent and answers 200. * So "absent from the listing" is a suspicion, not a verdict. This probe is * the verdict, and it is the only thing allowed to raise a warning. * * @return array|null array('url','headers','body') or null when unsupported. */ public static function inference_probe_request($provider, $api_key, $model) { $api_key = (string) $api_key; $model = (string) $model; if ($api_key === '' || $model === '') { return null; } $msg = array(array('role' => 'user', 'content' => 'hi')); switch ($provider) { case 'openai': return array( 'url' => 'https://api.openai.com/v1/chat/completions', 'headers' => array('Authorization' => 'Bearer ' . $api_key, 'Content-Type' => 'application/json'), 'body' => array('model' => $model, 'messages' => $msg, 'max_completion_tokens' => 1), ); case 'xai': return array( 'url' => 'https://api.x.ai/v1/chat/completions', 'headers' => array('Authorization' => 'Bearer ' . $api_key, 'Content-Type' => 'application/json'), 'body' => array('model' => $model, 'messages' => $msg, 'max_tokens' => 1), ); case 'claude': return array( 'url' => 'https://api.anthropic.com/v1/messages', 'headers' => array( 'x-api-key' => $api_key, 'anthropic-version' => '2023-06-01', 'Content-Type' => 'application/json', ), 'body' => array('model' => $model, 'max_tokens' => 1, 'messages' => $msg), ); case 'gemini': return array( 'url' => 'https://generativelanguage.googleapis.com/v1beta/models/' . rawurlencode($model) . ':generateContent', 'headers' => array('x-goog-api-key' => $api_key, 'Content-Type' => 'application/json'), 'body' => array( 'contents' => array(array('parts' => array(array('text' => 'hi')))), 'generationConfig' => array('maxOutputTokens' => 1), ), ); } return null; } /** * Flat allowlist of every chat model id across all providers. Used by the * sanitize() flow in class-mxchat-admin.php and the AJAX autosave validator * in admin/class-ajax-handler.php so a new model id only needs editing in * the catalog above. */ public static function chat_model_ids() { $ids = array(); foreach (self::chat_models() as $provider) { foreach ($provider['models'] as $id => $_unused) { $ids[] = $id; } } return $ids; } /** * Flat allowlist of every embedding model id across all providers. */ public static function embedding_model_ids() { $ids = array(); foreach (self::embedding_models() as $provider) { foreach ($provider['models'] as $id => $_unused) { $ids[] = $id; } } return $ids; } /** * Deprecation-aware picker label for a catalog entry (plan e46b8f). * * @param array $entry Model entry from chat_models(). * @return string|null Label to show (with a "retires " suffix when a * future 'deprecated_on' is set), or null when the * retirement date has passed and the entry must be * hidden from NEW selections. */ private static function picker_label($entry) { if (empty($entry['deprecated_on'])) { return $entry['label']; } $ts = strtotime($entry['deprecated_on'] . ' 00:00:00 UTC'); if ($ts !== false && time() >= $ts) { return null; // retired — hide from new selections } return sprintf( /* translators: 1: model label, 2: retirement date */ __('%1$s — retires %2$s', 'mxchat'), $entry['label'], date_i18n(get_option('date_format', 'F j, Y'), $ts) ); } /** * Settings-dropdown-shaped chat model map keyed by group label, which is * what mxchat_model_callback() in class-mxchat-admin.php renders as * . Each model maps to its label-with-description string. * * Retired models (past 'deprecated_on') are hidden — EXCEPT the install's * currently saved model, which is re-injected with a "retired" marker so * the settings screen never renders a blank select (plan e46b8f). * * @param string $current_model Optionally, the install's saved model id. * @param array|null $provider_slugs Optional allowlist of provider slugs * (catalog keys, e.g. 'openai', 'claude'). * null = every provider. Lets a consumer * that can only dispatch to some providers * (the content generator — plan ccb751) * derive its picker from the same catalog. */ public static function settings_dropdown_groups($current_model = '', $provider_slugs = null) { $out = array(); foreach (self::chat_models() as $provider_slug => $provider) { if (is_array($provider_slugs) && !in_array($provider_slug, $provider_slugs, true)) { continue; } $out[$provider['label']] = array(); foreach ($provider['models'] as $id => $entry) { $label = self::picker_label($entry); if ($label === null && $id !== $current_model) { continue; } if ($label === null) { /* translators: %s: model label */ $label = sprintf(__('%s — retired by provider, choose a replacement', 'mxchat'), $entry['label']); } $out[$provider['label']][$id] = $label; } } return $out; } /** * Shape used by js/mxchat-admin.js's modal picker grid: provider slug => * array of { value, label, description }. Mirrors the original inline * `models = {…}` literal in js/mxchat-admin.js so the modal can read from * a localized PHP source without changing its render path. */ public static function js_picker_shape() { $out = array(); foreach (self::chat_models() as $provider_slug => $provider) { $out[$provider_slug] = array(); foreach ($provider['models'] as $id => $entry) { $label = self::picker_label($entry); if ($label === null) { continue; // retired — never offered as a new selection } $out[$provider_slug][] = array( 'value' => $id, 'label' => $label, 'description' => $entry['description'], ); } } return $out; } /** * Shape used by the Onboarding wizard's JS catalog (per-provider * `{ label, chatModels, embeddingModels, hasKey, requiresKeyToLoadModels }`). * `hasKey` is filled in by the caller after consulting mxchat_options. */ public static function onboarding_js_catalog() { $chat = self::chat_models(); $emb = self::embedding_models(); $out = array(); $all_slugs = array_unique(array_merge(array_keys($chat), array_keys($emb))); foreach ($all_slugs as $slug) { $chat_models = array(); if (isset($chat[$slug])) { foreach ($chat[$slug]['models'] as $id => $entry) { $label = self::picker_label($entry); if ($label === null) { continue; // retired — not offered during onboarding } $chat_models[$id] = $label; } } $embed_models = array(); if (isset($emb[$slug])) { foreach ($emb[$slug]['models'] as $id => $entry) { $embed_models[$id] = $entry['label']; } } $label = self::provider_label($slug); $requires_key = isset($chat[$slug]) ? !empty($chat[$slug]['requires_key_to_load_models']) : false; $disable_onb = isset($chat[$slug]) ? !empty($chat[$slug]['disable_in_onboarding']) : false; $out[$slug] = array( 'label' => $label, 'chatModels' => $chat_models, 'embeddingModels' => $embed_models, 'requiresKeyToLoadModels' => $requires_key, 'disableInOnboarding' => ($disable_onb || $requires_key), 'keyOption' => self::key_option_for_provider($slug), ); } return $out; } /** * Whether the given chat model can do model-driven function calling (tool use). * * Used by the native function-calling loop (plan-mxchat-20260617-a41dee) to * decide whether to offer tools to the model and to gate the admin toggle. * * Design: every modern flagship across the catalog's providers supports tool * calling, so the DEFAULT is "capable" for any model whose id matches a known * provider family prefix (or the OpenRouter meta). This deliberately avoids a * per-model allowlist that would go stale the moment the catalog gains a model * (the same drift trap documented in CLAUDE.md's model-registry checklist). * A model can be force-excluded via the $no_tools list, and the whole verdict * is overridable through the `mxchat_model_supports_tools` filter. * * @param string $model_id Chat model id (e.g. 'gpt-5.5', 'claude-opus-4-8') or * the 'openrouter' meta selector. * @return bool */ public static function supports_tools($model_id) { $model_id = (string) $model_id; // The OpenRouter meta resolves to a per-account sub-model at request time; // most routable models support tools, so default it capable (the loop // no-ops gracefully if the chosen sub-model rejects a tools array). if ($model_id === 'openrouter') { return (bool) apply_filters('mxchat_model_supports_tools', true, $model_id); } // Known tool-capable provider family prefixes across the catalog. $capable_prefixes = array( 'gpt-', 'o1-', 'o3-', 'o4-', // OpenAI 'claude-', // Anthropic 'gemini-', // Google 'grok-', 'xai-', // xAI 'deepseek-', // DeepSeek (OpenAI-compatible tools) 'custom-', // OpenAI-compatible custom endpoints ); // Explicit exclusions for any catalog model known NOT to support tools. // Empty today — kept as the drift-safe escape hatch. $no_tools = array(); $supported = false; if (!in_array($model_id, $no_tools, true)) { foreach ($capable_prefixes as $prefix) { if (strpos($model_id, $prefix) === 0) { $supported = true; break; } } } return (bool) apply_filters('mxchat_model_supports_tools', $supported, $model_id); } /** * The catalog's CURRENT recommended model for image analysis, per provider. * * plan-mxchat-20260717-eb371e: add-ons that need "a vision-capable model" * (mxchat-vision's analysis paths, mxchat-admin-chat's in-admin image * analysis) previously froze a model id at their own call sites — and those * pins silently aged out (mxchat-vision was still sending gpt-4o). This is * the ONE place that opinion lives now; add-ons ask, with their old pin * demoted to a fallback for when mxchat-basic is absent or too old. * * Choices are deliberate: * - openai: gpt-5.6-sol — the current flagship, vision-capable; OpenAI's * stated replacement for the gpt-5.x-chat-latest aliases retired * 2026-08-10 (which this map previously pinned; plan e46b8f). * - claude: the current Haiku — image analysis wants fast + cheap; Haiku * is vision-capable and the catalog's newest Haiku generation. * - xai: grok-4.6, the current flagship. Promoted 2026-08-14 (plan * e1aa4a) from grok-4-0709, which was two generations behind and no * longer appears in xAI's /v1/models listing. Image input is CONFIRMED, * not assumed: a 32x32 fixture probe on 2026-08-14 returned the correct * colour on grok-4.6 / 4.5 / 4.3 and both 4.20 variants (all HTTP 200). * grok-4.3 is the cheaper vision option (roughly half the image-token * price, 1M context) if cost ever outweighs capability here. * - gemini: current stable Flash — vision-capable, no per-image surcharge. * * NOT covered: image-GENERATION models (Gemini Imagen / Flash Image etc.). * Those are a different capability the catalog does not model; add-ons that * pin one (mxchat-vision's editing endpoints) are correctly pinned. * * @param string $provider 'openai' | 'claude' | 'xai' | 'gemini'. * @return string Model id, or '' for an unknown provider (caller keeps its * fallback). Overridable via mxchat_default_vision_model. */ public static function default_vision_model($provider) { $map = array( 'openai' => 'gpt-5.6-sol', 'claude' => 'claude-haiku-4-5-20251001', 'xai' => 'grok-4.6', 'gemini' => 'gemini-3.8-flash', ); $model = isset($map[$provider]) ? $map[$provider] : ''; return (string) apply_filters('mxchat_default_vision_model', $model, $provider); } /** * Per-model, per-surface reasoning_effort value. * * plan-mxchat-20260714-dcb71c: the single source of truth for the * reasoning_effort ladder that was previously hand-maintained in FIVE * parallel places (integrator streaming/non-streaming chat, integrator * web-search Responses, content-generator, mxchat-advanced-content * class-ai-client + class-content-calendar). Adding a model to * chat_models() and updating THIS method now propagates the routing to * every surface — no more drift-on-every-model-add. * * The mapping is DELIBERATELY per-context because the surfaces genuinely * differ today (they are NOT a single clamped ladder): * - 'chat' — /v1/chat/completions main chat. gpt-5.5 and gpt-5.4 use * 'none'; the *-chat-latest / 5.4-mini / 5.4-nano / * gpt-5.2 ids send NO reasoning_effort; the original * gpt-5 / gpt-5-mini / gpt-5-nano send 'minimal' (their * generation requires it — 25b972 probe). Unknown future * gpt-5* ids send NOTHING (allowlist fall-through). * - 'content' — content generation (content-generator, AC ai-client, * AC content-calendar). gpt-5.5 / gpt-5.4 use 'low'; * gpt-5.4-mini / gpt-5.4-nano use 'none' (they never * supported 'minimal' — 25b972); gpt-5.3-chat-latest and * the original gpt-5 / -mini / -nano use 'minimal'; only * gpt-5.2 and gpt-5.1-chat-latest send nothing. This is * an ALLOWLIST (5621dd safe-default): a future gpt-5* id * not listed gets NO param rather than a guessed value * that might 400. * - 'websearch' — /v1/responses web-search. Only the gpt-5.6 trio, 5.5, * 5.4, and 5.1-2025-11-13 send 'low'; every other gpt-5 id * (including gpt-5 / mini / nano / chat-latest) sends * nothing. * * @param string $model Chat model id. * @param string $context 'chat' | 'content' | 'websearch'. Defaults to 'chat'. * @return string|null 'none'|'low'|'minimal' to send, or null to omit the * reasoning_effort / reasoning.effort param entirely. */ public static function reasoning_effort_for($model, $context = 'chat') { $model = (string) $model; // plan a9774b: every Gemini 3.x output price is "including thinking // tokens" and the model thinks at its default level unless told // otherwise — measured on this site's own support prompt, 71% of // billed output on 3.8 Flash was hidden thinking, with indistinguishable // replies at 'low' (0 hidden tokens). So the chat surface sends 'low': // as reasoning_effort on the OpenAI-compatible endpoint, and as // generationConfig.thinkingConfig.thinkingLevel on the native one (see // gemini_thinking_config()). Flash-Lite does not think and returns 400 // on any reasoning parameter — it must send nothing. if (strpos($model, 'gemini-3') === 0) { if (strpos($model, 'flash-lite') !== false) { return null; } return ($context === 'chat') ? 'low' : null; } // Only the modern OpenAI family (gpt-5*, gpt-6*) carries // reasoning_effort. Everything else omits. plan 45c22d widened this // from a literal gpt-5 prefix: a gpt-6 id used to fall through all // three maps below and be sent legacy parameters. if (!self::is_modern_openai($model)) { return null; } switch ($context) { case 'websearch': // integrator web-search (Responses API). Explicit allowlist, no // "else" — anything not listed omits reasoning entirely. $map = array( 'gpt-6-astra' => 'low', // plan 45c22d, probed 2026-09-27 on /v1/responses 'gpt-6-sol' => 'low', 'gpt-6-luna' => 'low', 'gpt-5.6-sol' => 'low', 'gpt-5.6-terra' => 'low', 'gpt-5.6-luna' => 'low', 'gpt-5.5' => 'low', 'gpt-5.4' => 'low', 'gpt-5.1-2025-11-13' => 'low', ); return isset($map[$model]) ? $map[$model] : null; case 'content': // content-generator + AC class-ai-client + AC content-calendar. // ALLOWLIST (5621dd): a gpt-5* id absent here sends nothing. // // plan-25b972 (probed live 2026-08-13): supported reasoning_effort // VALUES are per-model. gpt-5.4-mini/nano never supported // 'minimal' (their floor is 'none' — the old 'minimal' entries // 400'd on every call); the original gpt-5/-mini/-nano still // REQUIRE 'minimal' and reject 'none'. Do not "harmonize" these // — the two generations genuinely differ. $map = array( // plan 45c22d (probed 2026-09-27): gpt-6-astra accepts only // low|medium|high; gpt-6-sol/-luna accept none|low|medium|high. // 'low' for all three, as for their 5.6 siblings. 'gpt-6-astra' => 'low', 'gpt-6-sol' => 'low', 'gpt-6-luna' => 'low', 'gpt-5.6-sol' => 'low', 'gpt-5.6-terra' => 'low', 'gpt-5.6-luna' => 'low', 'gpt-5.5' => 'low', 'gpt-5.4' => 'low', 'gpt-5.1-2025-11-13' => 'low', 'gpt-5.4-mini' => 'none', 'gpt-5.4-nano' => 'none', 'gpt-5.3-chat-latest' => 'minimal', 'gpt-5' => 'minimal', 'gpt-5-mini' => 'minimal', 'gpt-5-nano' => 'minimal', ); return isset($map[$model]) ? $map[$model] : null; case 'chat': default: // integrator streaming + non-streaming main chat. // gpt-5.1/5.3-chat-latest stay listed here (and in 'content') // even though the catalog no longer offers them: bot-level / // add-on-saved ids that miss the e46b8f migration must keep // routing correctly until every surface is swept. $no_reasoning = array('gpt-5.2', 'gpt-5.1-chat-latest', 'gpt-5.3-chat-latest', 'gpt-5.4-mini', 'gpt-5.4-nano'); if (in_array($model, $no_reasoning, true)) { return null; } if ($model === 'gpt-5.1-2025-11-13') { return 'low'; } if ($model === 'gpt-5.5') { return 'none'; } if ($model === 'gpt-5.4') { return 'none'; } // plan 45c22d (probed 2026-09-27): gpt-6-astra accepts only // low|medium|high; -sol/-luna accept none|low|medium|high. // 'low' for all three, as for their 5.6 siblings. if (in_array($model, array('gpt-6-astra', 'gpt-6-sol', 'gpt-6-luna'), true)) { return 'low'; } if (in_array($model, array('gpt-5.6-sol', 'gpt-5.6-terra', 'gpt-5.6-luna'), true)) { return 'low'; } // plan-25b972: the original gpt-5 generation REQUIRES 'minimal' // (probed 2026-08-13: 'none' is rejected for these three). if (in_array($model, array('gpt-5', 'gpt-5-mini', 'gpt-5-nano'), true)) { return 'minimal'; } // plan-25b972: allowlist fall-through (5621dd philosophy). An // unknown future gpt-5* id sends NO param — a guessed value is // exactly how the 5.4-mini/nano 400s shipped. return null; } } /** * Whether a NON-DEFAULT temperature value may be sent to this model. * * plan-mxchat-20260714-dcb71c: consolidates the scattered temperature * omissions. Two families reject a custom temperature: * - gpt-5* — accept only the default (1); any other value 400s. Callers * that send temperature=1 (the main chat path) need not consult this — * 1 is the default and is accepted. Callers that send a NON-default * value (content-generator 0.7, AC 0.7) gate on this and omit for gpt-5. * - The Anthropic flagships that removed the temperature param entirely * (claude-opus-4-7 / 4-8, claude-fable-5, claude-sonnet-5) — 400 if sent * at all. This is the list the integrator's mxchat_claude_omits_temperature() * and content-generator's call_claude() have hand-maintained. * * @param string $model Chat model id. * @return bool true if a custom temperature may be sent. */ /** * generationConfig.thinkingConfig for the native Gemini endpoint, or null * to send none (plan a9774b). Derived from reasoning_effort_for() so the * two Gemini paths — native generateContent and the OpenAI-compatible * endpoint — always agree on the thinking level. Overridable via * mxchat_gemini_thinking_level (return '' to send none). * * @param string $model Chat model id. * @return array|null e.g. array('thinkingLevel' => 'low'). */ public static function gemini_thinking_config($model) { $level = self::reasoning_effort_for($model, 'chat'); if (strpos((string) $model, 'gemini-') !== 0) { return null; } $level = apply_filters('mxchat_gemini_thinking_level', $level, $model); if (!is_string($level) || $level === '' || !in_array($level, array('minimal', 'low', 'medium', 'high'), true)) { return null; } return array('thinkingLevel' => $level); } public static function supports_temperature($model) { $model = (string) $model; // gpt-5 / gpt-6 families accept only the default temperature (1). if (self::is_modern_openai($model)) { return false; } // Anthropic flagships that reject the temperature param outright. // TRAP: exact-match list — every NEW Anthropic flagship (4.7 and newer) // must be added here or it 400s on every chat message while looking // fine in the picker. A claude-opus-* prefix rule would wrongly catch // claude-opus-4-6 and older, which still accept temperature. $claude_no_temp = array('claude-opus-5', 'claude-opus-4-7', 'claude-opus-4-8', 'claude-fable-5', 'claude-sonnet-5'); if (in_array($model, $claude_no_temp, true)) { return false; } return true; } /** * The completion-token parameter key an OpenAI /v1/chat/completions call * must use for this model. gpt-5* require 'max_completion_tokens'; the * legacy 'max_tokens' 400s. All other OpenAI-compatible providers * (xAI, DeepSeek, custom) use 'max_tokens'. * * plan-mxchat-20260714-dcb71c: centralizes the `$is_gpt5 ? * 'max_completion_tokens' : 'max_tokens'` switch duplicated in * content-generator and AC class-ai-client. * * @param string $model Chat model id. * @return string 'max_completion_tokens' | 'max_tokens' */ public static function openai_token_param($model) { return self::is_modern_openai($model) ? 'max_completion_tokens' : 'max_tokens'; } /** * plan 45c22d: ONE predicate for "a modern OpenAI chat model" — the * generations that take max_completion_tokens, only the default * temperature, and reasoning_effort. Every gate that used to test the * literal prefix gpt-5 routes through here, so the next generation is a * one-line change instead of nine edits. Before this, gpt-6-* ids fell * through all nine and were sent legacy parameters: a 400 on the first * message while looking fine in the picker. * * The frozen partial-upgrade fallbacks in the integrator, the content * generator and the editor assistant carry the same two-prefix test * inline (they exist for the window where this class is not loaded). * * @param string $model Chat model id. * @return bool */ public static function is_modern_openai($model) { $model = (string) $model; return strpos($model, 'gpt-5') === 0 || strpos($model, 'gpt-6') === 0; } } ```