# Tchavi > Tchavi is an AI API gateway for Africa. One API key and one credit balance give > access to 130 AI models (GPT-6, Claude, Gemini, DeepSeek, Mistral, Llama, Kling, > Veo, Nano Banana, Whisper and more) through an OpenAI-compatible API, paid with > mobile money (MTN MoMo, Orange Money, M-Pesa, Airtel Money, Moov Money) in > 12 African countries. No bank card, no KYC, no subscription. ## Key facts - What it is: a proxy in front of OpenAI, Anthropic, Google, DeepSeek, Mistral and other providers. Requests are passed through unchanged. For text, image and audio only usage metadata is logged, never the content; video jobs keep their prompt, and generated videos are deleted after 24 hours. - Base URL: `https://tchavi.com/api/v1` — drop-in for the OpenAI SDK by changing `baseURL` and the API key (keys start with `sk-tch-`). - Endpoints: chat completions (streaming supported), images, audio (Whisper transcription, text-to-speech), video generation and music generation (async jobs), embeddings, `GET /v1/models`. - Payment: mobile money only, through PawaPay. No Visa/Mastercard needed. Credits are added as soon as the operator confirms the payment. - Prices: credits sold at a flat rate, 0,33 F CFA per credit (9 000 credits for 3 000 F). Minimum top-up 1 000 F. Purchased credits never expire. First paid top-up adds 150 bonus credits. New accounts in the countries we can take payment from receive free trial credits when they verify their email (capped daily, and the offer can stop). - Billing rule: text models are billed per token (input and output at their own rates), rounded up once per request to 0.001 credit, with no minimum per request; images per image; TTS per character; transcription per started minute (Whisper) or per second of audio (Scribe); video per second. Credits carry three decimals. - Compared with paying providers directly: Tchavi is not cheaper per token across the board. What you get is one account, one key and one balance for 100+ models, top-ups by mobile money, credits that never expire, and no international card needed. Where a price lands against the provider's list price depends on the model: models without a discount carry a Tchavi margin above the list price, while some discounted models (marked on their page and in the creator tables) can come out at or below it. The exact price of every model is listed below. In the CFA franc zone there is no foreign-exchange fee (outside it, a 2 % conversion margin applies). - Rate limits by account level: 10 requests/min before any top-up, 30 after the first top-up, 120 at 10 000 F cumulative, 300 at 50 000 F cumulative. - Languages: French and English site and support (support@tchavi.com). Company pages: https://www.linkedin.com/company/tchavi and https://x.com/tchaviai ## Mobile money coverage (live configuration, read 2026-11-17) - Benin: Moov Money, MTN MoMo - Cameroon: MTN MoMo - Congo: Airtel Money, MTN MoMo - Côte d'Ivoire: MTN MoMo, Orange Money - DR Congo: Airtel Money, Orange Money, M-Pesa - Gabon: Airtel Money, Moov Money - Kenya: M-Pesa - Rwanda: Airtel Money, MTN MoMo - Senegal: Free Money, Orange Money - Sierra Leone: Orange Money - Uganda: Airtel Money, MTN MoMo - Zambia: MTN MoMo, Zamtel Kwacha ## Core pages - [Home](https://tchavi.com/en): Product overview. - [Models](https://tchavi.com/en/models): Catalogue of every model with capabilities and prices. - [Pricing](https://tchavi.com/en/pricing): Credits and how billing works. - [Documentation](https://tchavi.com/en/docs): API reference and integration guides. - [Blog](https://tchavi.com/en/blog): Guides and price comparisons for building with AI from Africa. - [Full text for agents](https://tchavi.com/llms-full.txt): this index plus every documentation page and article in one file. ## Documentation - [Quick Start](https://tchavi.com/en/docs/quickstart): Get your first AI response in 3 steps. ([markdown](https://tchavi.com/en/docs/quickstart/llms.txt)) - [Authentication](https://tchavi.com/en/docs/authentication): Authenticate every request with your API key. ([markdown](https://tchavi.com/en/docs/authentication/llms.txt)) - [Base URL](https://tchavi.com/en/docs/base-url): Point any OpenAI-compatible client at the Tchavi API. ([markdown](https://tchavi.com/en/docs/base-url/llms.txt)) - [Core concepts](https://tchavi.com/en/docs/principles): Tchavi is a secure gateway that gives you access to multiple AI providers through one API key and one credit balance. ([markdown](https://tchavi.com/en/docs/principles/llms.txt)) - [Streaming](https://tchavi.com/en/docs/streaming): Receive responses token-by-token as Server-Sent Events by setting stream: true. ([markdown](https://tchavi.com/en/docs/streaming/llms.txt)) - [Coding agents](https://tchavi.com/en/docs/coding-agents): Run Claude Code, OpenCode, Deep Code, Codex or Cursor on Tchavi credits, paid with mobile money. ([markdown](https://tchavi.com/en/docs/coding-agents/llms.txt)) - [Claude Code](https://tchavi.com/en/docs/coding-agents/claude-code): Point Claude Code at Tchavi with two environment variables and run it on credits topped up by mobile money. ([markdown](https://tchavi.com/en/docs/coding-agents/claude-code/llms.txt)) - [OpenCode](https://tchavi.com/en/docs/coding-agents/opencode): Add Tchavi to OpenCode as a custom OpenAI-compatible provider and pay for it with mobile money. ([markdown](https://tchavi.com/en/docs/coding-agents/opencode/llms.txt)) - [Deep Code](https://tchavi.com/en/docs/coding-agents/deep-code): Run DeepSeek's Deep Code agent on Tchavi by setting BASE_URL, and pay in FCFA by mobile money. ([markdown](https://tchavi.com/en/docs/coding-agents/deep-code/llms.txt)) - [Codex](https://tchavi.com/en/docs/coding-agents/codex): Add Tchavi as a model provider in ~/.codex/config.toml and run Codex on mobile-money credits. ([markdown](https://tchavi.com/en/docs/coding-agents/codex/llms.txt)) - [Cursor](https://tchavi.com/en/docs/coding-agents/cursor): Override Cursor's OpenAI base URL to route chat through Tchavi — and what that path does not cover. ([markdown](https://tchavi.com/en/docs/coding-agents/cursor/llms.txt)) - [FAQ](https://tchavi.com/en/docs/faq): Quick answers to the most common questions about using Tchavi. ([markdown](https://tchavi.com/en/docs/faq/llms.txt)) - [Chat Completions](https://tchavi.com/en/docs/chat-completions): Generate chat completions from a list of messages — the primary endpoint for text generation with all supported models. ([markdown](https://tchavi.com/en/docs/chat-completions/llms.txt)) - [Image Generation](https://tchavi.com/en/docs/image-generation): Generate and edit images from text prompts across all supported image models. ([markdown](https://tchavi.com/en/docs/image-generation/llms.txt)) - [Audio](https://tchavi.com/en/docs/audio): Text-to-speech and transcription (Whisper) endpoints for generating audio from text and converting audio files to text. ([markdown](https://tchavi.com/en/docs/audio/llms.txt)) - [Video Generation](https://tchavi.com/en/docs/video-generation): Generate videos from text or images asynchronously, with optional audio, resolution and aspect-ratio control. ([markdown](https://tchavi.com/en/docs/video-generation/llms.txt)) - [Music Generation](https://tchavi.com/en/docs/music-generation): Generate a song from a description, or re-sing a track you already have, as an asynchronous job billed per track. ([markdown](https://tchavi.com/en/docs/music-generation/llms.txt)) - [Embeddings](https://tchavi.com/en/docs/embeddings): Convert text into numeric vectors for semantic search, clustering, recommendations, and RAG pipelines. ([markdown](https://tchavi.com/en/docs/embeddings/llms.txt)) - [Decisions](https://tchavi.com/en/docs/decisions): Ask a model a fixed set of questions about a piece of text and get back typed answers with probabilities, instead of prose you have to parse. ([markdown](https://tchavi.com/en/docs/decisions/llms.txt)) - [Async Jobs](https://tchavi.com/en/docs/jobs): Track the status of asynchronous operations — video and music generation — via the jobs endpoints. ([markdown](https://tchavi.com/en/docs/jobs/llms.txt)) - [Models](https://tchavi.com/en/docs/models): Access 100+ AI models from OpenAI, Anthropic, Google and more through a single API, organized into budget groups. ([markdown](https://tchavi.com/en/docs/models/llms.txt)) - [Credits & Billing](https://tchavi.com/en/docs/credits): Tchavi uses a credit-based billing system — each API request consumes credits based on the model used and the number of tokens processed. ([markdown](https://tchavi.com/en/docs/credits/llms.txt)) - [Mobile money payments: countries and operators](https://tchavi.com/en/docs/payments): Pay for Tchavi credits with mobile money (MTN MoMo, Orange Money, M-Pesa, Airtel Money, Moov Money, Free Money, Zamtel), no bank card needed: the countries, operators and currencies accepted. ([markdown](https://tchavi.com/en/docs/payments/llms.txt)) - [Rate Limits](https://tchavi.com/en/docs/rate-limits): Rate limits depend on your account level, which is unlocked automatically based on your lifetime spend on the platform. ([markdown](https://tchavi.com/en/docs/rate-limits/llms.txt)) - [Error Handling](https://tchavi.com/en/docs/errors): Tchavi returns standard HTTP status codes with a structured JSON error body. ([markdown](https://tchavi.com/en/docs/errors/llms.txt)) - [SDKs](https://tchavi.com/en/docs/sdks): Use the official @tchavi/sdk, the OpenAI SDK as a drop-in, or plain cURL to call the API. ([markdown](https://tchavi.com/en/docs/sdks/llms.txt)) ## Blog - [Claude Fable 5.2 API: First Call, Cost and 4 Traps](https://tchavi.com/en/blog/claude-fable-5-2-api-first-call-four-refusals) (EN, 2026-11-10): Claude Fable 5.2 on Tchavi: a first call, what one answer costs (70 to 140 credits, 12% off) and four traps: three refused requests, one ignored parameter. ([markdown](https://tchavi.com/en/blog/claude-fable-5-2-api-first-call-four-refusals/llms.txt)) - [Claude Fable 5.2 API : premier appel, prix et 4 pièges](https://tchavi.com/fr/blog/claude-fable-5-2-api-premier-appel-quatre-refus) (FR, 2026-11-10): Claude Fable 5.2 sur Tchavi : premier appel, coût d'une réponse (23,5 à 46,7 F, remise de 12 %) et quatre pièges : trois requêtes refusées, un paramètre ignoré. ([markdown](https://tchavi.com/fr/blog/claude-fable-5-2-api-premier-appel-quatre-refus/llms.txt)) - [Claude Fable 5.2 ou Opus 5.5 : lequel choisir, chiffres à l’appui](https://tchavi.com/fr/blog/claude-fable-5-2-ou-opus-5-5) (FR, 2026-11-10): Claude Fable 5.2 ou Opus 5.5 ? Même interface, 1,76 fois le prix aujourd’hui. Anthropic, un test indépendant et notre calcul : xhigh coûte 2 fois low. ([markdown](https://tchavi.com/fr/blog/claude-fable-5-2-ou-opus-5-5/llms.txt)) - [Claude Fable 5.2 vs Opus 5.5: Which One to Pick, With Numbers](https://tchavi.com/en/blog/claude-fable-5-2-vs-opus-5-5) (EN, 2026-11-10): Claude Fable 5.2 vs Opus 5.5: same interface, 1.76x the price today. Anthropic’s numbers, one independent test, our calculation: xhigh costs 2x low. ([markdown](https://tchavi.com/en/blog/claude-fable-5-2-vs-opus-5-5/llms.txt)) - [AI Voice API for French and Swahili, No Credit Card Needed](https://tchavi.com/en/blog/ai-voice-french-swahili-api-no-credit-card) (EN, 2026-11-06): ElevenLabs, MiniMax Speech and Scribe v2 landed on Tchavi in late October. Seven models tested on the same French sentence, a Swahili round trip, and the model to start with depending on your budget. ([markdown](https://tchavi.com/en/blog/ai-voice-french-swahili-api-no-credit-card/llms.txt)) - [Voix IA en français et en swahili par API, sans carte bancaire](https://tchavi.com/fr/blog/voix-ia-francais-swahili-api-sans-carte) (FR, 2026-11-06): ElevenLabs, MiniMax Speech et Scribe v2 sont sur Tchavi depuis fin octobre. Sept modèles testés sur la même phrase française, un aller-retour en swahili, et le modèle par lequel démarrer selon ton budget. ([markdown](https://tchavi.com/fr/blog/voix-ia-francais-swahili-api-sans-carte/llms.txt)) - [Combien coûte vraiment une chanson générée par IA ?](https://tchavi.com/fr/blog/combien-coute-une-chanson-generee-par-ia) (FR, 2026-11-05): Le prix réel d’une chanson ou d’une reprise générées par IA sur Tchavi : tarif fixe au morceau, ce que 3 000 F permettent, et le remboursement en cas d’échec. ([markdown](https://tchavi.com/fr/blog/combien-coute-une-chanson-generee-par-ia/llms.txt)) - [Créer un jingle IA pour ta boutique : prompt et prix](https://tchavi.com/fr/blog/creer-un-jingle-pour-votre-boutique-avec-l-ia) (FR, 2026-11-05): Génère un jingle publicitaire avec l'IA pour ta boutique, ton spot radio ou tes réseaux sociaux : prompt, paroles, option instrumentale et prix en FCFA. ([markdown](https://tchavi.com/fr/blog/creer-un-jingle-pour-votre-boutique-avec-l-ia/llms.txt)) - [Make an AI Jingle for Your Shop: Prompt, Lyrics and Price](https://tchavi.com/en/blog/make-a-jingle-for-your-shop-with-ai) (EN, 2026-11-05): Generate an AI jingle for your shop, a radio spot or a social post: how to write the prompt and lyrics, the instrumental option, and what it costs. ([markdown](https://tchavi.com/en/blog/make-a-jingle-for-your-shop-with-ai/llms.txt)) - [What an AI-Generated Song Actually Costs](https://tchavi.com/en/blog/what-an-ai-generated-song-costs) (EN, 2026-11-05): The real price of an AI-generated song or cover on Tchavi: a flat rate per track, what a 3,000 F top-up buys, and what happens when a job fails. ([markdown](https://tchavi.com/en/blog/what-an-ai-generated-song-costs/llms.txt)) - [Generate a Song with an AI API: From Prompt to MP3](https://tchavi.com/en/blog/generate-a-song-with-an-ai-api) (EN, 2026-11-04): A full tutorial for generating a song through the Tchavi API: prompt vs lyrics, the async job, polling, errors caught before billing, and real cost. ([markdown](https://tchavi.com/en/blog/generate-a-song-with-an-ai-api/llms.txt)) - [Générer une chanson avec une API IA : du prompt au MP3](https://tchavi.com/fr/blog/generer-une-chanson-avec-une-api-ia) (FR, 2026-11-04): Tutoriel complet pour générer une chanson avec l'API Tchavi : prompt contre paroles, job asynchrone, polling, erreurs avant facturation et coût réel. ([markdown](https://tchavi.com/fr/blog/generer-une-chanson-avec-une-api-ia/llms.txt)) - [Kimi K3.1 Pro est disponible sur Tchavi](https://tchavi.com/fr/blog/kimi-k3-1-pro-disponible-tchavi) (FR, 2026-10-23): Kimi K3.1 Pro arrive sur Tchavi : 1M de contexte, images et outils, à moitié prix d'Opus 5.5. Benchmarks vérifiés et limites mesurées. ([markdown](https://tchavi.com/fr/blog/kimi-k3-1-pro-disponible-tchavi/llms.txt)) - [Kimi K3.1 Pro is now on Tchavi](https://tchavi.com/en/blog/kimi-k3-1-pro-now-on-tchavi) (EN, 2026-10-23): Kimi K3.1 Pro is on Tchavi: 1M context, images and tools, at half the price of Opus 5.5. Verified benchmarks and limits measured on day one. ([markdown](https://tchavi.com/en/blog/kimi-k3-1-pro-now-on-tchavi/llms.txt)) - [Cost vs quality: find the cheapest AI model that clears your bar](https://tchavi.com/en/blog/cost-vs-quality-choosing-the-right-model) (EN, 2026-10-21): A 5-step method to choose an AI model: 15 scored cases, a pass bar, the cost of 1,000 tasks in KES. Node.js script, 4 models measured in October 2026. ([markdown](https://tchavi.com/en/blog/cost-vs-quality-choosing-the-right-model/llms.txt)) - [Coût vs qualité : le modèle le moins cher qui tient ta barre](https://tchavi.com/fr/blog/cout-qualite-choisir-le-bon-modele) (FR, 2026-10-21): Une méthode en 5 étapes pour choisir un modèle d'IA : 15 cas notés, une barre de réussite, le coût des 1 000 tâches en FCFA. Script Node.js, 4 modèles mesurés. ([markdown](https://tchavi.com/fr/blog/cout-qualite-choisir-le-bon-modele/llms.txt)) - [Ollama with a cloud API: Open WebUI plus Tchavi, paid by M-Pesa](https://tchavi.com/en/blog/ollama-tchavi-local-and-cloud-models) (EN, 2026-10-21): Ollama only serves its own models. Add Tchavi in Open WebUI or in your code for large cloud models, paid with mobile money: measurements and code. ([markdown](https://tchavi.com/en/blog/ollama-tchavi-local-and-cloud-models/llms.txt)) - [Ollama et Tchavi : modèles locaux et grands modèles sans carte](https://tchavi.com/fr/blog/ollama-tchavi-modeles-locaux-et-cloud) (FR, 2026-10-21): Ollama ne sert que ses propres modèles. Ajoute Tchavi dans Open WebUI ou dans ton code pour les grands modèles, payés en mobile money : mesures et code. ([markdown](https://tchavi.com/fr/blog/ollama-tchavi-modeles-locaux-et-cloud/llms.txt)) - [Pay for the OpenAI and Claude API with M-Pesa, no credit card](https://tchavi.com/en/blog/pay-openai-claude-api-mpesa-mobile-money) (EN, updated 2026-11-04): Card declined by OpenAI or Anthropic? Pay for the GPT and Claude API with M-Pesa or mobile money in Kenya, Uganda, Rwanda and Zambia, from KES 250. ([markdown](https://tchavi.com/en/blog/pay-openai-claude-api-mpesa-mobile-money/llms.txt)) - [API keys with a spend limit and an expiry date on Tchavi](https://tchavi.com/en/blog/api-keys-spend-limit-expiry) (EN, 2026-10-20): Every Tchavi API key can now have an expiry date and a spend limit in credits: how it works, the 403 error and what to watch for. ([markdown](https://tchavi.com/en/blog/api-keys-spend-limit-expiry/llms.txt)) - [Claude Code sans carte bancaire : le payer en mobile money](https://tchavi.com/fr/blog/claude-code-sans-carte-bancaire-mobile-money) (FR, 2026-10-20): Utiliser Claude Code sans carte Visa : la configuration en trois étapes, une clé plafonnée pour tenir ton budget, et le vrai coût d'une session en FCFA. ([markdown](https://tchavi.com/fr/blog/claude-code-sans-carte-bancaire-mobile-money/llms.txt)) - [Claude Code without a credit card: pay with M-Pesa or MoMo](https://tchavi.com/en/blog/claude-code-without-credit-card-mobile-money) (EN, 2026-10-20): Use Claude Code without a credit card or Anthropic subscription: a three-step setup, a spend-capped key, and what a real task cost in credits and KES. ([markdown](https://tchavi.com/en/blog/claude-code-without-credit-card-mobile-money/llms.txt)) - [Clés API plafonnées : limite de dépense et expiration](https://tchavi.com/fr/blog/cles-api-plafonnees-limite-depense-expiration) (FR, 2026-10-20): Chaque clé API Tchavi peut avoir une date d'expiration et une limite de dépense en crédits : fonctionnement, erreur 403 et points à surveiller. ([markdown](https://tchavi.com/fr/blog/cles-api-plafonnees-limite-depense-expiration/llms.txt)) - [DeepSeek V4.1 Flash, Mistral Large 4 and three more models](https://tchavi.com/en/blog/deepseek-v4-1-flash-mistral-large-4-new-models) (EN, updated 2026-11-02): DeepSeek V4.1 Flash, Mistral Large 4, Command A+, Gemini 3.5 Flash-Lite and MiMo-V2.6-Flash are on Tchavi: what each is for, and prices per million tokens. ([markdown](https://tchavi.com/en/blog/deepseek-v4-1-flash-mistral-large-4-new-models/llms.txt)) - [DeepSeek V4.1 Flash, Mistral Large 4 et trois autres modèles](https://tchavi.com/fr/blog/deepseek-v4-1-flash-mistral-large-4-nouveaux-modeles) (FR, updated 2026-11-02): DeepSeek V4.1 Flash, Mistral Large 4, Command A+, Gemini 3.5 Flash-Lite et MiMo-V2.6-Flash sont sur Tchavi : usages et prix par million de tokens. ([markdown](https://tchavi.com/fr/blog/deepseek-v4-1-flash-mistral-large-4-nouveaux-modeles/llms.txt)) - [Meilleurs modèles vidéo IA en 2026 : prix en FCFA par seconde](https://tchavi.com/fr/blog/meilleurs-modeles-video-ia-2026-prix-fcfa) (FR, 2026-10-19): Wan 3, Veo 3.1, Kling V3, Seedance 2.5 : quel modèle vidéo IA choisir en 2026, ce que chacun sait faire et son prix en FCFA par seconde, payé en mobile money. ([markdown](https://tchavi.com/fr/blog/meilleurs-modeles-video-ia-2026-prix-fcfa/llms.txt)) - [AI Image Generation and Editing API for Product Photos](https://tchavi.com/en/blog/ai-image-generation-editing-api-product-photos) (EN, updated 2026-11-04): Generate a product photo with an image API, then edit it from a reference image (white background, price tag). Node and Python code, costs and errors measured. ([markdown](https://tchavi.com/en/blog/ai-image-generation-editing-api-product-photos/llms.txt)) - [Générer et retoucher des images produit par l'API (Node, Python)](https://tchavi.com/fr/blog/api-generation-retouche-images-produit) (FR, updated 2026-11-04): Génère une photo produit par l'API, puis retouche-la depuis une image de référence (fond blanc, prix). Code Node et Python, coûts et erreurs mesurés. ([markdown](https://tchavi.com/fr/blog/api-generation-retouche-images-produit/llms.txt)) - [New image models: GPT Image 2.5, FLUX 3, Grok Imagine](https://tchavi.com/en/blog/new-image-models-gpt-image-2-5-flux-3-grok) (EN, 2026-10-18): GPT Image 2.5 Sunburst, FLUX 3 and Grok Imagine Image 2 arrive on Tchavi: what each is for, sizes, prices in credits and KES, and a call you can copy. ([markdown](https://tchavi.com/en/blog/new-image-models-gpt-image-2-5-flux-3-grok/llms.txt)) - [New video models: Wan 3, Seedance 2.5, Grok Imagine Lite](https://tchavi.com/en/blog/new-video-models-wan-3-seedance-2-5-grok) (EN, 2026-10-18): Wan 3, Seedance 2.5 and Grok Imagine Video 1.5 Lite arrive on Tchavi: durations, resolutions, audio, price per second and a call you can copy. ([markdown](https://tchavi.com/en/blog/new-video-models-wan-3-seedance-2-5-grok/llms.txt)) - [Nouveaux modèles image : GPT Image 2.5, FLUX 3, Grok Imagine](https://tchavi.com/fr/blog/nouveaux-modeles-image-gpt-image-2-5-flux-3-grok) (FR, 2026-10-18): GPT Image 2.5 Sunburst, FLUX 3 et Grok Imagine Image 2 arrivent sur Tchavi : à quoi sert chacun, formats, prix en crédits et en F, et un appel prêt à copier. ([markdown](https://tchavi.com/fr/blog/nouveaux-modeles-image-gpt-image-2-5-flux-3-grok/llms.txt)) - [Nouveaux modèles vidéo : Wan 3, Seedance 2.5, Grok Imagine Lite](https://tchavi.com/fr/blog/nouveaux-modeles-video-wan-3-seedance-2-5-grok) (FR, 2026-10-18): Wan 3, Seedance 2.5 et Grok Imagine Video 1.5 Lite arrivent sur Tchavi : durées, résolutions, son, prix par seconde et un appel prêt à copier. ([markdown](https://tchavi.com/fr/blog/nouveaux-modeles-video-wan-3-seedance-2-5-grok/llms.txt)) - [Tool calling in Node.js: one loop that works with any model](https://tchavi.com/en/blog/tool-calling-one-loop-any-model) (EN, updated 2026-11-02): Write one OpenAI-format tool calling loop and switch models by changing a string. Node.js code, loop guards, and what 4 models did in October 2026. ([markdown](https://tchavi.com/en/blog/tool-calling-one-loop-any-model/llms.txt)) - [Tool calling : une seule boucle, tous les modèles (Node.js)](https://tchavi.com/fr/blog/tool-calling-une-boucle-tous-les-modeles) (FR, updated 2026-11-02): Écris une boucle de tool calling au format OpenAI et change de modèle avec une chaîne. Code Node.js, garde-fous, et ce que 4 modèles ont fait en octobre 2026. ([markdown](https://tchavi.com/fr/blog/tool-calling-une-boucle-tous-les-modeles/llms.txt)) - [Jev ou un LLM : quand un modèle de décision suffit](https://tchavi.com/fr/blog/jev-ou-llm-quand-utiliser-un-modele-de-decision) (FR, 2026-10-16): Jev contre gpt-6-luna sur 40 messages WhatsApp de boutique : exactitude, erreurs par classe, latence et coût pour 1 000 classements, mesurés. ([markdown](https://tchavi.com/fr/blog/jev-ou-llm-quand-utiliser-un-modele-de-decision/llms.txt)) - [Jev vs LLM: when a decision model is enough](https://tchavi.com/en/blog/jev-vs-llm-when-to-use-a-decision-model) (EN, 2026-10-16): Jev vs gpt-6-luna on 40 shop WhatsApp messages: accuracy, per-class errors, latency and the cost of 1,000 classifications, all measured. ([markdown](https://tchavi.com/en/blog/jev-vs-llm-when-to-use-a-decision-model/llms.txt)) - [Quelle IA comprend le wolof ? Test de Whisper en 4 langues](https://tchavi.com/fr/blog/notes-vocales-whatsapp-ia-wolof-swahili) (FR, updated 2026-11-02): Quelle IA comprend le wolof, le swahili ou le lingala ? Test de Whisper sur 7 phrases lues (octobre 2026), résultats mot à mot et prix en FCFA. ([markdown](https://tchavi.com/fr/blog/notes-vocales-whatsapp-ia-wolof-swahili/llms.txt)) - [Vision par API : lire un reçu mobile money avec un LLM (JSON)](https://tchavi.com/fr/blog/vision-api-lire-recu-image-llm) (FR, updated 2026-11-04): Envoie la photo d'un reçu mobile money ou d'une facture à un LLM et récupère du JSON. Base64 ou URL, 3 modèles comparés, crédits et temps mesurés. ([markdown](https://tchavi.com/fr/blog/vision-api-lire-recu-image-llm/llms.txt)) - [Vision API: Read a Mobile Money Receipt with an LLM (JSON)](https://tchavi.com/en/blog/vision-api-read-receipts-images-llm) (EN, updated 2026-11-04): Send a photo of a mobile money receipt or an invoice to an LLM and get JSON back. Base64 or URL, 3 models compared, credits and timings measured. ([markdown](https://tchavi.com/en/blog/vision-api-read-receipts-images-llm/llms.txt)) - [Swahili speech-to-text: Whisper tested in 4 African languages](https://tchavi.com/en/blog/whatsapp-voice-notes-ai-swahili-transcription) (EN, updated 2026-11-02): Does AI transcribe Swahili? Whisper tested on 7 read sentences in Swahili, French, Lingala and Wolof (October 2026): word errors and cost in KES. ([markdown](https://tchavi.com/en/blog/whatsapp-voice-notes-ai-swahili-transcription/llms.txt)) - [One API for text, image, video and audio](https://tchavi.com/en/blog/one-api-text-image-video-audio) (EN, 2026-10-14): One Tchavi key and one base URL cover chat, images, voice, transcription, embeddings, video and Jev. Every call, measured in credits. ([markdown](https://tchavi.com/en/blog/one-api-text-image-video-audio/llms.txt)) - [Une seule API pour le texte, l'image, la vidéo et l'audio](https://tchavi.com/fr/blog/une-seule-api-texte-image-video-audio) (FR, 2026-10-14): Une clé Tchavi et une URL de base couvrent le chat, l'image, la voix, la transcription, les embeddings, la vidéo et Jev. Les appels, mesurés en crédits. ([markdown](https://tchavi.com/fr/blog/une-seule-api-texte-image-video-audio/llms.txt)) - [Bot de support IA : modèle économique d'abord, puis escalade](https://tchavi.com/fr/blog/bot-support-modele-economique-escalade) (FR, 2026-10-13): Un bot de support WhatsApp qui répond avec un petit modèle et n'appelle un modèle plus fort que si besoin. Code Node.js, 20 messages mesurés, coût en FCFA. ([markdown](https://tchavi.com/fr/blog/bot-support-modele-economique-escalade/llms.txt)) - [Support bot with a cheap AI model first, then escalation](https://tchavi.com/en/blog/support-bot-cheap-model-first-escalation) (EN, 2026-10-13): A WhatsApp support bot that answers with a small model and calls a stronger one only when needed. Node.js code, 20 messages measured, cost in KES. ([markdown](https://tchavi.com/en/blog/support-bot-cheap-model-first-escalation/llms.txt)) - [Jev : le modèle de décision expliqué aux développeurs](https://tchavi.com/fr/blog/jev-modele-de-decision-explique) (FR, 2026-10-12): Jev répond à une question par des probabilités typées, pas par du texte : modération, intention, preuve de paiement. Appels réels, seuils, coût en crédits. ([markdown](https://tchavi.com/fr/blog/jev-modele-de-decision-explique/llms.txt)) - [Synthèse vocale (TTS) par l'API : une voix française en 5 minutes](https://tchavi.com/fr/blog/synthese-vocale-tts-api-5-minutes) (FR, updated 2026-11-06): Donne une voix française à ton bot WhatsApp ou ton app avec l'API de synthèse vocale : cURL, Node, Python, SDK OpenAI, voix, instructions et coût au caractère. ([markdown](https://tchavi.com/fr/blog/synthese-vocale-tts-api-5-minutes/llms.txt)) - [Text-to-Speech API in 5 Minutes: cURL, Node, Python, OpenAI SDK](https://tchavi.com/en/blog/text-to-speech-api-5-minutes) (EN, updated 2026-11-06): Give your WhatsApp bot or app a voice with a text-to-speech API: cURL, Node, Python, the OpenAI SDK, voices, instructions and measured cost per character. ([markdown](https://tchavi.com/en/blog/text-to-speech-api-5-minutes/llms.txt)) - [What is Jev? The decision model explained for developers](https://tchavi.com/en/blog/what-is-jev-decision-model) (EN, 2026-10-12): Jev answers a question with typed probabilities, not text: moderation, intent, proof of payment. Real calls, thresholds and cost in credits. ([markdown](https://tchavi.com/en/blog/what-is-jev-decision-model/llms.txt)) - [Passerelle d'API IA : à quoi ça sert, et quand en avoir besoin](https://tchavi.com/fr/blog/passerelle-api-ia-a-quoi-ca-sert) (FR, 2026-10-08): Une passerelle d'API IA donne une clé et une URL pour plusieurs fournisseurs de modèles. Ce qu'elle fait, ce qu'elle ne fait pas, et quand aller en direct. ([markdown](https://tchavi.com/fr/blog/passerelle-api-ia-a-quoi-ca-sert/llms.txt)) - [What is an AI API gateway, and when do you need one?](https://tchavi.com/en/blog/what-is-an-ai-api-gateway) (EN, 2026-10-08): An AI API gateway gives you one key and one URL for several model providers. What it does, what it does not, and when to go direct. ([markdown](https://tchavi.com/en/blog/what-is-an-ai-api-gateway/llms.txt)) - [Claude Sonnet 5.5 est disponible sur Tchavi](https://tchavi.com/fr/blog/claude-sonnet-5-5-disponible-tchavi) (FR, 2026-10-05): Claude Sonnet 5.5 est ouvert sur Tchavi : contexte de 1M tokens, sortie de 128K, images et PDF. Ce qu'il sait faire, sa limite sur les outils. ([markdown](https://tchavi.com/fr/blog/claude-sonnet-5-5-disponible-tchavi/llms.txt)) - [Claude Sonnet 5.5 is now available on Tchavi](https://tchavi.com/en/blog/claude-sonnet-5-5-now-on-tchavi) (EN, 2026-10-05): Claude Sonnet 5.5 is open on Tchavi: 1M-token context, 128K output, images and PDFs. What it does well, and its limit on forced tool calls. ([markdown](https://tchavi.com/en/blog/claude-sonnet-5-5-now-on-tchavi/llms.txt)) - [Chatbot WhatsApp IA en Node.js : guide et coût en FCFA](https://tchavi.com/fr/blog/chatbot-whatsapp-ia-nodejs-afrique) (FR, updated 2026-10-22): Chatbot WhatsApp IA au Bénin, Sénégal, Côte d'Ivoire : API WhatsApp Cloud, webhook Node.js, réponses GPT-6 Luna ou Claude, déploiement VPS et coût réel en FCFA. ([markdown](https://tchavi.com/fr/blog/chatbot-whatsapp-ia-nodejs-afrique/llms.txt)) - [Prix des API IA en FCFA (2026) : GPT-6, Claude, DeepSeek](https://tchavi.com/fr/blog/prix-api-ia-fcfa-2026) (FR, updated 2026-11-02): GPT-6, Claude, DeepSeek, Gemini : prix en FCFA par million de tokens, par image et par seconde de vidéo au 22 octobre 2026, avec 3 budgets types. ([markdown](https://tchavi.com/fr/blog/prix-api-ia-fcfa-2026/llms.txt)) - [Payer l'API OpenAI ou Claude sans carte Visa, en mobile money](https://tchavi.com/fr/blog/payer-api-ia-sans-carte-visa) (FR, updated 2026-11-04): Carte refusée par OpenAI ou Anthropic ? Paie l'API de GPT-6, Claude et Mistral en mobile money (MTN MoMo, Orange Money, M-Pesa), à partir de 1 000 F. ([markdown](https://tchavi.com/fr/blog/payer-api-ia-sans-carte-visa/llms.txt)) ## APIs by creator - [OpenAI API pricing](https://tchavi.com/en/m/openai): 31 OpenAI models, from 36.7 F CFA to 6,233 F CFA / 1M input tokens (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API OpenAI : prix en F CFA](https://tchavi.com/fr/m/openai) : 31 modèles OpenAI, de 36,7 F CFA à 6 233 F CFA / 1M tokens en entrée (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [Gemini API pricing](https://tchavi.com/en/m/google): 15 Google models, from 194 F CFA to 1,467 F CFA / 1M input tokens (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API Gemini : prix en F CFA](https://tchavi.com/fr/m/google) : 15 modèles Google, de 194 F CFA à 1 467 F CFA / 1M tokens en entrée (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [Claude API pricing](https://tchavi.com/en/m/anthropic): 13 Anthropic models, from 73.3 F CFA to 5,499 F CFA / 1M input tokens (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API Claude : prix en F CFA](https://tchavi.com/fr/m/anthropic) : 13 modèles Anthropic, de 73,3 F CFA à 5 499 F CFA / 1M tokens en entrée (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [Qwen API pricing](https://tchavi.com/en/m/alibaba): 12 Alibaba models, from 22 F CFA to 2,933 F CFA / 1M input tokens (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API Qwen : prix en F CFA](https://tchavi.com/fr/m/alibaba) : 12 modèles Alibaba, de 22 F CFA à 2 933 F CFA / 1M tokens en entrée (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [DeepSeek API pricing](https://tchavi.com/en/m/deepseek): 3 DeepSeek models, from 104 F CFA to 870 F CFA / 1M input tokens (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API DeepSeek : prix en F CFA](https://tchavi.com/fr/m/deepseek) : 3 modèles DeepSeek, de 104 F CFA à 870 F CFA / 1M tokens en entrée (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [Seedance API pricing](https://tchavi.com/en/m/bytedance): 7 ByteDance models, from 11.7 F CFA to 28.8 F CFA per image (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API Seedance : prix en F CFA](https://tchavi.com/fr/m/bytedance) : 7 modèles ByteDance, de 11,7 F CFA à 28,8 F CFA par image (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [MiniMax API pricing](https://tchavi.com/en/m/minimax): 10 MiniMax models, from 220 F CFA to 220 F CFA / 1M input tokens (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API MiniMax : prix en F CFA](https://tchavi.com/fr/m/minimax) : 10 modèles MiniMax, de 220 F CFA à 220 F CFA / 1M tokens en entrée (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [Mistral API pricing](https://tchavi.com/en/m/mistral): 4 Mistral AI models, from 110 F CFA to 995 F CFA / 1M input tokens (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API Mistral : prix en F CFA](https://tchavi.com/fr/m/mistral) : 4 modèles Mistral AI, de 110 F CFA à 995 F CFA / 1M tokens en entrée (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [Grok API pricing](https://tchavi.com/en/m/xai): 4 xAI models, from 1,467 F CFA to 1,467 F CFA / 1M input tokens (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API Grok : prix en F CFA](https://tchavi.com/fr/m/xai) : 4 modèles xAI, de 1 467 F CFA à 1 467 F CFA / 1M tokens en entrée (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [Kimi API pricing](https://tchavi.com/en/m/moonshot-ai): 3 Moonshot AI models, from 367 F CFA to 2,200 F CFA / 1M input tokens (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API Kimi : prix en F CFA](https://tchavi.com/fr/m/moonshot-ai) : 3 modèles Moonshot AI, de 367 F CFA à 2 200 F CFA / 1M tokens en entrée (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [Kling API pricing](https://tchavi.com/en/m/kuaishou): 3 Kuaishou models, from 92.3 F CFA to 123 F CFA per second of video (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API Kling : prix en F CFA](https://tchavi.com/fr/m/kuaishou) : 3 modèles Kuaishou, de 92,3 F CFA à 123 F CFA par seconde de vidéo (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [FLUX API pricing](https://tchavi.com/en/m/black-forest-labs): 3 Black Forest Labs models, from 17.7 F CFA to 36.8 F CFA per image (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API FLUX : prix en F CFA](https://tchavi.com/fr/m/black-forest-labs) : 3 modèles Black Forest Labs, de 17,7 F CFA à 36,8 F CFA par image (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA - [elevenlabs API pricing](https://tchavi.com/en/m/elevenlabs): 11 elevenlabs models, from 2.68 F CFA to 58.5 F CFA per minute of audio (as of 2026-11-17); one Tchavi key, OpenAI-compatible, mobile money top-up from 1,000 F CFA - [API elevenlabs : prix en F CFA](https://tchavi.com/fr/m/elevenlabs) : 11 modèles elevenlabs, de 2,68 F CFA à 58,5 F CFA par minute d'audio (au 2026-11-17) ; une clé Tchavi, compatible SDK OpenAI, recharge Mobile Money à partir de 1 000 F CFA ## Models and prices (credits; 1 credit ≈ 0,33 F CFA) ### chat - [DeepSeek V4 Flash](https://tchavi.com/en/m/deepseek/deepseek-v4-flash) — id `deepseek-v4-flash`, by deepseek: 0.336 / 0.615 credits per 1K tokens in / out (≈ 112 F / 205 F per 1M tokens) - [DeepSeek V4.1 Flash](https://tchavi.com/en/m/deepseek/deepseek-v4.1-flash) — id `deepseek-v4.1-flash`, by deepseek: 0.33 / 1.317 credits per 1K tokens in / out (≈ 110 F / 439 F per 1M tokens) - [Free Router](https://tchavi.com/en/m/tchavi/free-router) — id `free-router`, by Tchavi: 0.01 / 0.01 credits per 1K tokens in / out (≈ 3 F / 3 F per 1M tokens) - [GLM 5.3 Flash](https://tchavi.com/en/m/z.ai/glm-5.3-flash) — id `glm-5.3-flash`, by Z.ai: 0.18 / 0.6 credits per 1K tokens in / out (≈ 60 F / 200 F per 1M tokens) - [GPT-4o Mini](https://tchavi.com/en/m/openai/gpt-4o-mini) — id `gpt-4o-mini`, by openai: 0.33 / 1.3 credits per 1K tokens in / out (≈ 110 F / 433 F per 1M tokens) - [GPT-5 Nano](https://tchavi.com/en/m/openai/gpt-5-nano) — id `gpt-5-nano`, by openai: 0.11 / 0.88 credits per 1K tokens in / out (≈ 37 F / 293 F per 1M tokens) - [GPT-5.4 Nano](https://tchavi.com/en/m/openai/gpt-5.4-nano) — id `gpt-5.4-nano`, by openai: 0.44 / 2.7 credits per 1K tokens in / out (≈ 147 F / 900 F per 1M tokens) - [GPT-6 Luna](https://tchavi.com/en/m/openai/gpt-6-luna) — id `gpt-6-luna`, by openai: 0.22 / 1.1 credits per 1K tokens in / out (≈ 73 F / 367 F per 1M tokens) - [Ling 3.0 Flash VL](https://tchavi.com/en/m/inclusionai/ling-3.0-flash-vl) — id `ling-3.0-flash-vl`, by InclusionAI: 0.132 / 0.396 credits per 1K tokens in / out (≈ 44 F / 132 F per 1M tokens) - [MiMo-V2.6-Flash](https://tchavi.com/en/m/xiaomi/mimo-v2.6-flash) — id `mimo-v2.6-flash`, by Xiaomi: 0.308 / 0.615 credits per 1K tokens in / out (≈ 103 F / 205 F per 1M tokens) - [Mistral Small](https://tchavi.com/en/m/mistral/mistral-small-latest) — id `mistral-small-latest`, by mistral: 0.33 / 1.3 credits per 1K tokens in / out (≈ 110 F / 433 F per 1M tokens) - [Qwen3.7 Flash](https://tchavi.com/en/m/alibaba/qwen3.7-flash) — id `qwen3.7-flash`, by Alibaba: 0.066 / 0.29 credits per 1K tokens in / out (≈ 22 F / 97 F per 1M tokens) - [Qwen3.8 Flash](https://tchavi.com/en/m/alibaba/qwen3.8-flash) — id `qwen3.8-flash`, by Alibaba: 0.33 / 1.04 credits per 1K tokens in / out (≈ 110 F / 347 F per 1M tokens) - [Qwen3.8 Omni Flash](https://tchavi.com/en/m/alibaba/qwen3.8-omni-flash) — id `qwen3.8-omni-flash`, by Alibaba: 0.348 / 1.089 credits per 1K tokens in / out (≈ 116 F / 363 F per 1M tokens) - [Claude Haiku](https://tchavi.com/en/m/anthropic/claude-haiku-4-5-20251001) — id `claude-haiku-4-5-20251001`, by anthropic: 2.2 / 11 credits per 1K tokens in / out (≈ 733 F / 3,667 F per 1M tokens) - [Claude Haiku 5.5](https://tchavi.com/en/m/anthropic/claude-haiku-5-5) — id `claude-haiku-5-5`, by anthropic: 0.22 / 1.1 credits per 1K tokens in / out (≈ 73 F / 367 F per 1M tokens) - [Command A+](https://tchavi.com/en/m/cohere/command-a-plus) — id `command-a-plus`, by Cohere: 0.659 / 3.293 credits per 1K tokens in / out (≈ 220 F / 1,098 F per 1M tokens) - [Gemini 3.1 Flash Lite](https://tchavi.com/en/m/google/gemini-3.1-flash-lite) — id `gemini-3.1-flash-lite`, by google: 0.581 / 3.482 credits per 1K tokens in / out (≈ 194 F / 1,161 F per 1M tokens) - [Gemini 3.5 Flash](https://tchavi.com/en/m/google/gemini-3.5-flash) — id `gemini-3.5-flash`, by google: 3.482 / 21.1 credits per 1K tokens in / out (≈ 1,161 F / 7,033 F per 1M tokens) - [Gemini 3.5 Flash-Lite](https://tchavi.com/en/m/google/gemini-3.5-flash-lite) — id `gemini-3.5-flash-lite`, by google: 0.696 / 5.79 credits per 1K tokens in / out (≈ 232 F / 1,930 F per 1M tokens) - [Gemini 3.8 Flash](https://tchavi.com/en/m/google/gemini-3.8-flash) — id `gemini-3.8-flash`, by google: 1.741 / 8.44 credits per 1K tokens in / out (≈ 580 F / 2,813 F per 1M tokens) - [GLM-5.3](https://tchavi.com/en/m/z.ai/glm-5.3) — id `glm-5.3`, by Z.ai: 2.2 / 6.8 credits per 1K tokens in / out (≈ 733 F / 2,267 F per 1M tokens) - [GPT-4.1 Mini](https://tchavi.com/en/m/openai/gpt-4.1-mini) — id `gpt-4.1-mini`, by openai: 0.88 / 3.5 credits per 1K tokens in / out (≈ 293 F / 1,167 F per 1M tokens) - [GPT-5 Mini](https://tchavi.com/en/m/openai/gpt-5-mini) — id `gpt-5-mini`, by openai: 0.55 / 4.4 credits per 1K tokens in / out (≈ 183 F / 1,467 F per 1M tokens) - [GPT-5.4 Mini](https://tchavi.com/en/m/openai/gpt-5.4-mini) — id `gpt-5.4-mini`, by openai: 1.65 / 10 credits per 1K tokens in / out (≈ 550 F / 3,333 F per 1M tokens) - [GPT-6.1 Terra](https://tchavi.com/en/m/openai/gpt-6.1-terra) — id `gpt-6.1-terra`, by openai: 2.195 / 10.975 credits per 1K tokens in / out (≈ 732 F / 3,658 F per 1M tokens) - [Grok 4.6](https://tchavi.com/en/m/xai/grok-4.6) — id `grok-4.6`, by xAI: 4.4 / 13.2 credits per 1K tokens in / out (≈ 1,467 F / 4,400 F per 1M tokens) - [Grok 4.7](https://tchavi.com/en/m/xai/grok-4.7) — id `grok-4.7`, by xAI: 4.4 / 13.2 credits per 1K tokens in / out (≈ 1,467 F / 4,400 F per 1M tokens) - [Kimi K2.5](https://tchavi.com/en/m/moonshot-ai/kimi-k2.5) — id `kimi-k2.5`, by Moonshot AI: 1.158 / 5.79 credits per 1K tokens in / out (≈ 386 F / 1,930 F per 1M tokens) - [Llama 3.3 70B](https://tchavi.com/en/m/meta/llama-3.3-70b) — id `llama-3.3-70b`, by Meta: 0.308 / 0.9 credits per 1K tokens in / out (≈ 103 F / 300 F per 1M tokens) - [MiMo-V2.6-Pro](https://tchavi.com/en/m/xiaomi/mimo-v2.6-pro) — id `mimo-v2.6-pro`, by Xiaomi: 0.96 / 1.91 credits per 1K tokens in / out (≈ 320 F / 637 F per 1M tokens) - [MiniMax M3](https://tchavi.com/en/m/minimax/minimax-m3) — id `minimax-m3`, by MiniMax: 0.66 / 2.6 credits per 1K tokens in / out (≈ 220 F / 867 F per 1M tokens) - [Mistral Large 4](https://tchavi.com/en/m/mistral/mistral-large-4) — id `mistral-large-4`, by mistral: 2.986 / 9.175 credits per 1K tokens in / out (≈ 995 F / 3,058 F per 1M tokens) - [Mistral Large](https://tchavi.com/en/m/mistral/mistral-large-latest) — id `mistral-large-latest`, by mistral: 1.1 / 3.3 credits per 1K tokens in / out (≈ 367 F / 1,100 F per 1M tokens) - [Nova 2 Lite](https://tchavi.com/en/m/amazon/nova-2-lite) — id `nova-2-lite`, by Amazon: 0.66 / 5.5 credits per 1K tokens in / out (≈ 220 F / 1,833 F per 1M tokens) - [Qwen 2.5 72B](https://tchavi.com/en/m/alibaba/qwen-2.5-72b) — id `qwen-2.5-72b`, by Alibaba: 0.8 / 0.88 credits per 1K tokens in / out (≈ 267 F / 293 F per 1M tokens) - [Qwen 2.5 Coder 32B](https://tchavi.com/en/m/alibaba/qwen-2.5-coder-32b) — id `qwen-2.5-coder-32b`, by Alibaba: 1.45 / 2.2 credits per 1K tokens in / out (≈ 483 F / 733 F per 1M tokens) - [Qwen 2.5 VL 72B](https://tchavi.com/en/m/alibaba/qwen-2.5-vl-72b) — id `qwen-2.5-vl-72b`, by Alibaba: 1.76 / 2.2 credits per 1K tokens in / out (≈ 587 F / 733 F per 1M tokens) - [Qwen3.8 Max](https://tchavi.com/en/m/alibaba/qwen3.8-max) — id `qwen3.8-max`, by Alibaba: 4.4 / 13 credits per 1K tokens in / out (≈ 1,467 F / 4,333 F per 1M tokens) - [Claude Opus](https://tchavi.com/en/m/anthropic/claude-opus-4-6) — id `claude-opus-4-6`, by anthropic: 11 / 55 credits per 1K tokens in / out (≈ 3,667 F / 18,333 F per 1M tokens) - [Claude Opus 4.7](https://tchavi.com/en/m/anthropic/claude-opus-4-7) — id `claude-opus-4-7`, by anthropic: 11 / 55 credits per 1K tokens in / out (≈ 3,667 F / 18,333 F per 1M tokens) - [Claude Opus 4.8](https://tchavi.com/en/m/anthropic/claude-opus-4-8) — id `claude-opus-4-8`, by anthropic: 11 / 55 credits per 1K tokens in / out (≈ 3,667 F / 18,333 F per 1M tokens) - [Claude Opus 5](https://tchavi.com/en/m/anthropic/claude-opus-5) — id `claude-opus-5`, by anthropic: 11 / 55 credits per 1K tokens in / out (≈ 3,667 F / 18,333 F per 1M tokens) - [Claude Opus 5.5](https://tchavi.com/en/m/anthropic/claude-opus-5-5) — id `claude-opus-5-5`, by anthropic: 8.8 / 44 credits per 1K tokens in / out (≈ 2,933 F / 14,667 F per 1M tokens) - [Claude Sonnet](https://tchavi.com/en/m/anthropic/claude-sonnet-4-5-20250929) — id `claude-sonnet-4-5-20250929`, by anthropic: 6.6 / 33 credits per 1K tokens in / out (≈ 2,200 F / 11,000 F per 1M tokens) - [Claude Sonnet](https://tchavi.com/en/m/anthropic/claude-sonnet-4-6) — id `claude-sonnet-4-6`, by anthropic: 6.6 / 33 credits per 1K tokens in / out (≈ 2,200 F / 11,000 F per 1M tokens) - [Claude Sonnet 5](https://tchavi.com/en/m/anthropic/claude-sonnet-5) — id `claude-sonnet-5`, by anthropic: 4.4 / 22 credits per 1K tokens in / out (≈ 1,467 F / 7,333 F per 1M tokens) - [Claude Sonnet 5.5](https://tchavi.com/en/m/anthropic/claude-sonnet-5-5) — id `claude-sonnet-5-5`, by anthropic: 4.4 / 22 credits per 1K tokens in / out (≈ 1,467 F / 7,333 F per 1M tokens) - [Command A](https://tchavi.com/en/m/cohere/command-a) — id `command-a`, by Cohere: 5.5 / 22 credits per 1K tokens in / out (≈ 1,833 F / 7,333 F per 1M tokens) - [DeepSeek V4 Pro](https://tchavi.com/en/m/deepseek/deepseek-v4-pro) — id `deepseek-v4-pro`, by deepseek: 2.9 / 8.7 credits per 1K tokens in / out (≈ 967 F / 2,900 F per 1M tokens) - [Gemini 3.1 Pro](https://tchavi.com/en/m/google/gemini-3.1-pro) — id `gemini-3.1-pro`, by google: 4.642 / 27.747 credits per 1K tokens in / out (≈ 1,547 F / 9,249 F per 1M tokens) - [Gemini 4 Argon](https://tchavi.com/en/m/google/gemini-4-argon) — id `gemini-4-argon`, by google: 8.8 / 44 credits per 1K tokens in / out (≈ 2,933 F / 14,667 F per 1M tokens) - [GPT-4.1](https://tchavi.com/en/m/openai/gpt-4.1) — id `gpt-4.1`, by openai: 4.4 / 17.6 credits per 1K tokens in / out (≈ 1,467 F / 5,867 F per 1M tokens) - [GPT-4o](https://tchavi.com/en/m/openai/gpt-4o) — id `gpt-4o`, by openai: 5.5 / 22 credits per 1K tokens in / out (≈ 1,833 F / 7,333 F per 1M tokens) - [GPT-5](https://tchavi.com/en/m/openai/gpt-5) — id `gpt-5`, by openai: 2.7 / 22 credits per 1K tokens in / out (≈ 900 F / 7,333 F per 1M tokens) - [GPT-5.1](https://tchavi.com/en/m/openai/gpt-5.1) — id `gpt-5.1`, by openai: 2.7 / 22 credits per 1K tokens in / out (≈ 900 F / 7,333 F per 1M tokens) - [GPT-5.2](https://tchavi.com/en/m/openai/gpt-5.2) — id `gpt-5.2`, by openai: 4.39 / 31 credits per 1K tokens in / out (≈ 1,463 F / 10,333 F per 1M tokens) - [GPT-5.4](https://tchavi.com/en/m/openai/gpt-5.4) — id `gpt-5.4`, by openai: 5.5 / 33 credits per 1K tokens in / out (≈ 1,833 F / 11,000 F per 1M tokens) - [GPT-5.5](https://tchavi.com/en/m/openai/gpt-5.5) — id `gpt-5.5`, by openai: 11 / 66 credits per 1K tokens in / out (≈ 3,667 F / 22,000 F per 1M tokens) - [GPT-6 Sol](https://tchavi.com/en/m/openai/gpt-6-sol) — id `gpt-6-sol`, by openai: 4.4 / 22 credits per 1K tokens in / out (≈ 1,467 F / 7,333 F per 1M tokens) - [GPT-6.1 Sol](https://tchavi.com/en/m/openai/gpt-6.1-sol) — id `gpt-6.1-sol`, by openai: 4.4 / 22 credits per 1K tokens in / out (≈ 1,467 F / 7,333 F per 1M tokens) - [Kimi K3](https://tchavi.com/en/m/moonshot-ai/kimi-k3) — id `kimi-k3`, by Moonshot AI: 6.6 / 32 credits per 1K tokens in / out (≈ 2,200 F / 10,667 F per 1M tokens) - [Kimi K3.1 Pro](https://tchavi.com/en/m/moonshot-ai/kimi-k3.1-pro) — id `kimi-k3.1-pro`, by Moonshot AI: 4.39 / 21.95 credits per 1K tokens in / out (≈ 1,463 F / 7,317 F per 1M tokens) - [o3](https://tchavi.com/en/m/openai/o3) — id `o3`, by openai: 4.4 / 17.6 credits per 1K tokens in / out (≈ 1,467 F / 5,867 F per 1M tokens) - [Qwen3.8 Max Prime](https://tchavi.com/en/m/alibaba/qwen3.8-max-prime) — id `qwen3.8-max-prime`, by Alibaba: 8.8 / 26 credits per 1K tokens in / out (≈ 2,933 F / 8,667 F per 1M tokens) - [Claude Fable 5.1](https://tchavi.com/en/m/anthropic/claude-fable-5-1) — id `claude-fable-5-1`, by anthropic: 22 / 110 credits per 1K tokens in / out (≈ 7,333 F / 36,667 F per 1M tokens) - [Claude Fable 5.2](https://tchavi.com/en/m/anthropic/claude-fable-5-2) — id `claude-fable-5-2`, by anthropic: 17.56 / 87.81 credits per 1K tokens in / out (≈ 5,853 F / 29,270 F per 1M tokens) - [GPT-6 Astra](https://tchavi.com/en/m/openai/gpt-6-astra) — id `gpt-6-astra`, by openai: 22 / 110 credits per 1K tokens in / out (≈ 7,333 F / 36,667 F per 1M tokens) ### audio - [Eleven Flash v2](https://tchavi.com/en/m/elevenlabs/eleven-flash-v2) — id `eleven-flash-v2`, by elevenlabs: 87.798 credits (≈ 29.3 F) per 1 000 characters - [Eleven Flash v2.5](https://tchavi.com/en/m/elevenlabs/eleven-flash-v2.5) — id `eleven-flash-v2.5`, by elevenlabs: 87.798 credits (≈ 29.3 F) per 1 000 characters - [Eleven Multilingual v2](https://tchavi.com/en/m/elevenlabs/eleven-multilingual-v2) — id `eleven-multilingual-v2`, by elevenlabs: 175.596 credits (≈ 58.5 F) per 1 000 characters - [Eleven Turbo v2](https://tchavi.com/en/m/elevenlabs/eleven-turbo-v2) — id `eleven-turbo-v2`, by elevenlabs: 87.798 credits (≈ 29.3 F) per 1 000 characters - [Eleven Turbo v2.5](https://tchavi.com/en/m/elevenlabs/eleven-turbo-v2.5) — id `eleven-turbo-v2.5`, by elevenlabs: 87.798 credits (≈ 29.3 F) per 1 000 characters - [Eleven v3](https://tchavi.com/en/m/elevenlabs/eleven-v3) — id `eleven-v3`, by elevenlabs: 175.596 credits (≈ 58.5 F) per 1 000 characters - [Eleven v3 Conversational](https://tchavi.com/en/m/elevenlabs/eleven-v3-conversational) — id `eleven-v3-conversational`, by elevenlabs: 87.798 credits (≈ 29.3 F) per 1 000 characters - [Eleven v4](https://tchavi.com/en/m/elevenlabs/eleven-v4) — id `eleven-v4`, by elevenlabs: 175.596 credits (≈ 58.5 F) per 1 000 characters - [Eleven v4 Turbo](https://tchavi.com/en/m/elevenlabs/eleven-v4-turbo) — id `eleven-v4-turbo`, by elevenlabs: 87.798 credits (≈ 29.3 F) per 1 000 characters - [Gemini 3.8 Flash-Lite TTS](https://tchavi.com/en/m/google/gemini-3.8-flash-lite-tts) — id `gemini-3.8-flash-lite-tts`, by google: 32 credits (≈ 10.7 F) per 1 000 characters - [MiniMax Speech 2.8 HD](https://tchavi.com/en/m/minimax/minimax-speech-2.8-hd) — id `minimax-speech-2.8-hd`, by MiniMax: 219.495 credits (≈ 73.2 F) per 1 000 characters - [MiniMax Speech 2.8 Turbo](https://tchavi.com/en/m/minimax/minimax-speech-2.8-turbo) — id `minimax-speech-2.8-turbo`, by MiniMax: 131.697 credits (≈ 43.9 F) per 1 000 characters - [Scribe v2](https://tchavi.com/en/m/elevenlabs/scribe-v2) — id `scribe-v2`, by elevenlabs: 8.049 credits (≈ 2.7 F) per minute of audio - [Scribe v2 Medical](https://tchavi.com/en/m/elevenlabs/scribe-v2-medical) — id `scribe-v2-medical`, by elevenlabs: 8.049 credits (≈ 2.7 F) per minute of audio - [Standard TTS](https://tchavi.com/en/m/openai/tts-1) — id `tts-1`, by openai: 32 credits (≈ 10.7 F) per 1 000 characters - [Voxtral 3 TTS](https://tchavi.com/en/m/mistral/voxtral-tts-3) — id `voxtral-tts-3`, by mistral: 29 credits (≈ 9.7 F) per 1 000 characters - [Whisper](https://tchavi.com/en/m/openai/whisper-1) — id `whisper-1`, by openai: 13 credits (≈ 4.3 F) per minute of audio - [TTS Mini (Steerable)](https://tchavi.com/en/m/openai/gpt-4o-mini-tts) — id `gpt-4o-mini-tts`, by openai: 40 credits (≈ 13.3 F) per 1 000 characters - [HD TTS](https://tchavi.com/en/m/openai/tts-1-hd) — id `tts-1-hd`, by openai: 64 credits (≈ 21.3 F) per 1 000 characters ### image - [GPT Image 1 Mini](https://tchavi.com/en/m/openai/gpt-image-1-mini) — id `gpt-image-1-mini`, by openai: 11 credits (≈ 4 F) per image - [FLUX.2 Pro](https://tchavi.com/en/m/black-forest-labs/flux-2-pro) — id `flux-2-pro`, by black-forest-labs: 65 credits (≈ 22 F) per image - [FLUX 3](https://tchavi.com/en/m/black-forest-labs/flux-3-image) — id `flux-3-image`, by black-forest-labs: 1K: 53 credits (≈ 18 F) per image, 1.5K: 77 credits (≈ 26 F) per image, 2K: 110 credits (≈ 37 F) per image, 4K: 667 credits (≈ 222 F) per image - [GPT Image 1.5](https://tchavi.com/en/m/openai/gpt-image-1.5) — id `gpt-image-1.5`, by openai: 28 credits (≈ 9 F) per image - [GPT Image 2](https://tchavi.com/en/m/openai/gpt-image-2) — id `gpt-image-2`, by openai: 13 credits (≈ 4 F) per image - [GPT Image 2.5 Sunburst](https://tchavi.com/en/m/openai/gpt-image-2.5-sunburst) — id `gpt-image-2.5-sunburst`, by openai: 16 credits (≈ 5 F) per image - [GPT Image 2.6 Flare](https://tchavi.com/en/m/openai/gpt-image-2.6-flare) — id `gpt-image-2.6-flare`, by openai: 16 credits (≈ 5 F) per image - [GPT Image 2.6 Sunburst](https://tchavi.com/en/m/openai/gpt-image-2.6-sunburst) — id `gpt-image-2.6-sunburst`, by openai: 16 credits (≈ 5 F) per image - [Grok Imagine Image 2](https://tchavi.com/en/m/xai/grok-imagine-image-2) — id `grok-imagine-image-2`, by xAI: 1K: 88 credits (≈ 29 F) per image, 2K: 88 credits (≈ 29 F) per image - [Nano Banana](https://tchavi.com/en/m/google/nano-banana) — id `nano-banana`, by google: 1K: 83 credits (≈ 28 F) per image - [Nano Banana Flash-Lite](https://tchavi.com/en/m/google/nano-banana-flash-lite) — id `nano-banana-flash-lite`, by google: 1K: 72 credits (≈ 24 F) per image, 2K: 107 credits (≈ 36 F) per image, 4K: 160 credits (≈ 53 F) per image - [Qwen Image 3](https://tchavi.com/en/m/alibaba/qwen-image-3) — id `qwen-image-3`, by Alibaba: 73 credits (≈ 24 F) per image - [Seedream 4.5](https://tchavi.com/en/m/bytedance/seedream-4.5) — id `seedream-4.5`, by bytedance: 85 credits (≈ 28 F) per image - [Seedream 5.0 Flash](https://tchavi.com/en/m/bytedance/seedream-5-flash) — id `seedream-5-flash`, by bytedance: 35 credits (≈ 12 F) per image - [Seedream 5.0 Lite](https://tchavi.com/en/m/bytedance/seedream-5-lite) — id `seedream-5-lite`, by bytedance: 75 credits (≈ 25 F) per image - [FLUX.2 Flex](https://tchavi.com/en/m/black-forest-labs/flux-2-flex) — id `flux-2-flex`, by black-forest-labs: 130 credits (≈ 43 F) per image - [Nano Banana 2](https://tchavi.com/en/m/google/nano-banana-2) — id `nano-banana-2`, by google: 0.5K: 100 credits (≈ 33 F) per image, 1K: 148 credits (≈ 49 F) per image, 2K: 214 credits (≈ 71 F) per image, 4K: 320 credits (≈ 107 F) per image - [Nano Banana 2.1](https://tchavi.com/en/m/google/nano-banana-2.1) — id `nano-banana-2.1`, by google: 1K: 74 credits (≈ 25 F) per image, 2K: 111 credits (≈ 37 F) per image, 4K: 249 credits (≈ 83 F) per image - [Qwen Image 3 Pro](https://tchavi.com/en/m/alibaba/qwen-image-3-pro) — id `qwen-image-3-pro`, by Alibaba: 1K: 95 credits (≈ 32 F) per image, 2K: 172 credits (≈ 57 F) per image - [Seedream 5.0 Pro](https://tchavi.com/en/m/bytedance/seedream-5-pro) — id `seedream-5-pro`, by bytedance: 1K: 96 credits (≈ 32 F) per image, 2K: 191 credits (≈ 64 F) per image - [Nano Banana Pro](https://tchavi.com/en/m/google/nano-banana-pro) — id `nano-banana-pro`, by google: 2K: 299.62 credits (≈ 100 F) per image, 4K: 607.68 credits (≈ 203 F) per image, 1K: 299.62 credits (≈ 100 F) per image ### decision - [Jev 1.13](https://tchavi.com/en/m/typesafe/jev-1.13) — id `jev-1.13`, by TypeSafe: 0.09 / 0 credits per 1K tokens in / out (≈ 30 F / 0 F per 1M tokens) ### video - [MiniMax H3 Lite](https://tchavi.com/en/m/minimax/minimax-h3-lite) — id `minimax-h3-lite`, by MiniMax: billed per second of generated video, see the model page for the per-resolution rate - [Grok Imagine Video 1.5 Lite](https://tchavi.com/en/m/xai/grok-imagine-video-1.5-lite) — id `grok-imagine-video-1.5-lite`, by xAI: billed per second of generated video, see the model page for the per-resolution rate - [Hailuo 2.3](https://tchavi.com/en/m/minimax/hailuo-2.3) — id `hailuo-2.3`, by MiniMax: billed per second of generated video, see the model page for the per-resolution rate - [Hailuo 3](https://tchavi.com/en/m/minimax/hailuo-3) — id `hailuo-3`, by MiniMax: billed per second of generated video, see the model page for the per-resolution rate - [Hailuo 3 Max](https://tchavi.com/en/m/minimax/hailuo-3-max) — id `hailuo-3-max`, by MiniMax: billed per second of generated video, see the model page for the per-resolution rate - [MiniMax H3 Highspeed](https://tchavi.com/en/m/minimax/minimax-h3-highspeed) — id `minimax-h3-highspeed`, by MiniMax: billed per second of generated video, see the model page for the per-resolution rate - [Seedance 2.0](https://tchavi.com/en/m/bytedance/seedance-2) — id `seedance-2`, by bytedance: billed per second of generated video, see the model page for the per-resolution rate - [Seedance 2.0 Fast](https://tchavi.com/en/m/bytedance/seedance-2-fast) — id `seedance-2-fast`, by bytedance: billed per second of generated video, see the model page for the per-resolution rate - [Seedance 2.5](https://tchavi.com/en/m/bytedance/seedance-2.5) — id `seedance-2.5`, by bytedance: billed per second of generated video, see the model page for the per-resolution rate - [Veo 3.1 Fast](https://tchavi.com/en/m/google/veo-3.1-fast) — id `veo-3.1-fast`, by google: billed per second of generated video, see the model page for the per-resolution rate - [Veo 3.1 Lite](https://tchavi.com/en/m/google/veo-3.1-lite) — id `veo-3.1-lite`, by google: billed per second of generated video, see the model page for the per-resolution rate - [Wan 2.6](https://tchavi.com/en/m/alibaba/wan-2.6) — id `wan-2.6`, by Alibaba: billed per second of generated video, see the model page for the per-resolution rate - [Wan 3.0](https://tchavi.com/en/m/alibaba/wan-3) — id `wan-3`, by Alibaba: billed per second of generated video, see the model page for the per-resolution rate - [Kling V3](https://tchavi.com/en/m/kuaishou/kling-v3) — id `kling-v3`, by Kuaishou: billed per second of generated video, see the model page for the per-resolution rate - [Kling V3 Motion Control](https://tchavi.com/en/m/kuaishou/kling-v3-motion-control) — id `kling-v3-motion-control`, by Kuaishou: billed per second of generated video, see the model page for the per-resolution rate - [Kling V4](https://tchavi.com/en/m/kuaishou/kling-v4) — id `kling-v4`, by Kuaishou: billed per second of generated video, see the model page for the per-resolution rate - [Veo 3.1](https://tchavi.com/en/m/google/veo-3.1) — id `veo-3.1`, by google: billed per second of generated video, see the model page for the per-resolution rate ### embedding - [Embedding Large](https://tchavi.com/en/m/openai/text-embedding-3-large) — id `text-embedding-3-large`, by openai: 0.29 / 0 credits per 1K tokens in / out (≈ 97 F / 0 F per 1M tokens) - [Embedding Small](https://tchavi.com/en/m/openai/text-embedding-3-small) — id `text-embedding-3-small`, by openai: 0.044 / 0 credits per 1K tokens in / out (≈ 15 F / 0 F per 1M tokens) ### music - [MiniMax Music 2.6](https://tchavi.com/en/m/minimax/minimax-music-2.6) — id `minimax-music-2.6`, by MiniMax: 329.242 credits (≈ 110 F) per generated track - [MiniMax Music Cover](https://tchavi.com/en/m/minimax/minimax-music-cover) — id `minimax-music-cover`, by MiniMax: 658.484 credits (≈ 219 F) per generated track