# -*- coding: utf-8 -*- """Single source of truth for selectable AI models, defaults and prices. Verified against the providers' official docs on 2026-10-08. resources/settings.xml `values=` lists must match MODELS; UsageTracker.get_pricing reads PRICES (via price_for). Prices are USD per 1M tokens (standard tier). """ import datetime MODELS = { 'openai': ['gpt-6-luna', 'gpt-6.1-sol', 'gpt-6-astra'], 'anthropic': ['claude-haiku-5-5', 'claude-sonnet-5-5', 'claude-opus-5-5'], 'google': ['gemini-3.8-flash', 'gemini-3.5-flash-lite', 'gemini-3.1-pro-preview'], 'mistral': ['mistral-medium-latest', 'mistral-large-4', 'mistral-small-latest', 'ministral-14b-latest', 'ministral-8b-latest', 'ministral-3b-latest'], 'groq': ['openai/gpt-oss-120b', 'openai/gpt-oss-20b', 'qwen/qwen3.8-27b'], } DEFAULT_MODEL = {provider: models[0] for provider, models in MODELS.items()} SETTING_ID = { 'openai': 'openai_model', 'anthropic': 'anthropic_model', 'google': 'google_model', 'mistral': 'mistral_model', 'groq': 'groq_model', } # Per model: input, cached_input (cache read; a model without it bills cached tokens like input), output, and optionally # cache_write price of tokens written to a prompt cache (Anthropic 5-minute write; else billed like input), # long_context {above, input, cached_input, output} the whole request is billed at these prices when its prompt is # longer than `above` tokens, # from {YYYY-MM-DD: {price overrides}} announced price changes, applied from that day on (see price_for). PRICES = { 'openai': { 'gpt-6-astra': {'input': 10.00, 'cached_input': 1.00, 'output': 50.00, 'long_context': {'above': 272_000, 'input': 20.00, 'cached_input': 2.00, 'output': 75.00}}, 'gpt-6.1-sol': {'input': 2.00, 'cached_input': 0.10, 'output': 10.00, 'long_context': {'above': 272_000, 'input': 4.00, 'cached_input': 0.20, 'output': 15.00}}, 'gpt-6-luna': {'input': 0.10, 'cached_input': 0.01, 'output': 0.50, 'long_context': {'above': 272_000, 'input': 0.20, 'cached_input': 0.02, 'output': 0.75}}, }, 'anthropic': { # Haiku 5.5 bills a prompt over 100K tokens at higher prices (the whole request); cache reads cost 0.1x input 'claude-haiku-5-5': {'input': 0.10, 'cached_input': 0.01, 'cache_write': 0.125, 'output': 0.50, 'long_context': {'above': 100_000, 'input': 0.50, 'cached_input': 0.05, 'cache_write': 0.625, 'output': 2.50}}, 'claude-sonnet-5-5': {'input': 2.00, 'cached_input': 0.10, 'cache_write': 2.50, 'output': 10.00}, 'claude-opus-5-5': {'input': 4.00, 'cached_input': 0.20, 'cache_write': 5.00, 'output': 20.00}, }, 'google': { 'gemini-3.8-flash': {'input': 0.75, 'cached_input': 0.075, 'output': 3.75, 'from': {'2027-01-01': {'input': 1.50, 'cached_input': 0.15, 'output': 7.50}}}, 'gemini-3.5-flash-lite': {'input': 0.30, 'cached_input': 0.03, 'output': 2.50}, 'gemini-3.1-pro-preview': {'input': 2.00, 'cached_input': 0.20, 'output': 12.00, 'long_context': {'above': 200_000, 'input': 4.00, 'cached_input': 0.40, 'output': 18.00}}, }, 'mistral': { 'mistral-medium-latest': {'input': 1.50, 'output': 7.50}, # public preview launch price; regular price is $1.36 / $0.14 / $4.18 'mistral-large-4': {'input': 0.68, 'cached_input': 0.07, 'output': 2.09}, 'mistral-small-latest': {'input': 0.15, 'output': 0.60}, 'ministral-14b-latest': {'input': 0.20, 'output': 0.20}, 'ministral-8b-latest': {'input': 0.15, 'output': 0.15}, 'ministral-3b-latest': {'input': 0.10, 'output': 0.10}, }, 'groq': { # prompt caching (automatic, 50% off cached input) exists for the gpt-oss models only 'openai/gpt-oss-120b': {'input': 0.15, 'cached_input': 0.075, 'output': 0.60}, 'openai/gpt-oss-20b': {'input': 0.075, 'cached_input': 0.0375, 'output': 0.30}, 'qwen/qwen3.8-27b': {'input': 0.80, 'output': 4.00}, }, } # Claude models that accept output_config.effort (they all think adaptively) CLAUDE_EFFORT_MODELS = ('claude-haiku-5-5', 'claude-sonnet-5-5', 'claude-opus-5-5') # ... and the ones whose refusals the API retries server-side (`fallbacks`); Haiku 5.5 has no server-side fallback CLAUDE_FALLBACK_MODELS = ('claude-sonnet-5-5', 'claude-opus-5-5') TTS_MODELS = { 'openai': ['gpt-4o-mini-tts', 'gpt-4o-mini-tts-2025-12-15', 'gpt-4o-mini-tts-2025-03-20'], 'elevenlabs': ['eleven_flash_v2_5', 'eleven_v4_turbo', 'eleven_v4', 'eleven_v3'], } # ElevenLabs models that interpret inline [audio tags] instead of reading them aloud ELEVENLABS_AUDIO_TAG_MODELS = ('eleven_v3', 'eleven_v4', 'eleven_v4_turbo') def price_for(provider, model, today=None): """Prices of a catalog model valid on `today` (price changes from PRICES[..]['from'] applied), or None if unknown.""" base = PRICES.get(provider, {}).get(model) if base is None: return None day = (today or datetime.date.today()).isoformat() price = {key: value for key, value in base.items() if key != 'from'} for start, change in sorted(base.get('from', {}).items()): if start <= day: price.update(change) return price def model_or_default(provider, model): """Return model if it is currently selectable, else the provider default.""" if model in MODELS.get(provider, ()): return model return DEFAULT_MODEL.get(provider, model) def migrate_model_settings(addon): """Replace stored models that are no longer offered with the provider default. Returns a list of (setting_id, old, new) changes. """ changes = [] for provider, setting_id in SETTING_ID.items(): stored = addon.getSetting(setting_id) if stored and stored not in MODELS[provider]: new = DEFAULT_MODEL[provider] addon.setSetting(setting_id, new) changes.append((setting_id, stored, new)) tts_stored = addon.getSetting('milionar_tts_model') if tts_stored and tts_stored not in TTS_MODELS['openai']: addon.setSetting('milionar_tts_model', TTS_MODELS['openai'][0]) changes.append(('milionar_tts_model', tts_stored, TTS_MODELS['openai'][0])) return changes