Chat Completion API Expansion: 15 New Models Across 6 Modalities

# Chat Completion API Expansion: 15 New Models Across 6 Modalities We've significantly expanded our Chat Completion API with **15 new production-ready models** spanning six different modalities - all accessible through the same unified endpoint you're already using. ## What Changed in Our API Our managed Chat Completion API now supports a much broader range of AI capabilities. Whether you need advanced reasoning, specialized code generation, audio transcription, image creation, or semantic search embeddings - it's all available through one endpoint with zero API key management. ### Previous Model Support Before this update, our Chat Completion API provided access to: - **Text Generation**: Anthropic Claude models (Opus, Sonnet), Meta LLAMA models, and 3 open-weight text models - **Code**: NousCoder provider for general-purpose code generation This gave developers a solid foundation, but we heard your feedback: you need specialized models for specific use cases. ## New Models Available Today We've added **15 specialized models** across six modalities: ### 🧠 Advanced Reasoning (NEW CATEGORY) Models that show their work and think step-by-step through complex problems: | Model | Best For | Plan Tier | |-------|----------|-----------| | **deepseek-r1** | Complex problem-solving with full chain-of-thought | Enterprise | | **deepseek-r1-distill-qwen-7b** | Efficient reasoning at 7B parameters | Professional+ | | **deepseek-r1-distill-llama-8b** | Llama-based reasoning optimization | Professional+ | **Use Cases:** Mathematical proofs, multi-step planning, code debugging, logical analysis ### 💻 Additional Code Models Purpose-built coding assistants trained on massive code repositories. These complement our existing NousCoder provider with specialized alternatives: | Model | Parameters | Plan Tier | |-------|------------|-----------| | **qwen-coder-7b** | 7B | Basic+ | | **qwen-coder-32b** | 32B | Professional+ | **Use Cases:** Code completion, refactoring, documentation generation, API design **Note:** NousCoder remains available for general-purpose code generation across all tiers. ### 🎙️ Audio Transcription (NEW CATEGORY) Speech-to-text powered by Whisper architecture: | Model | Accuracy | Plan Tier | |-------|----------|-----------| | **whisper-small** | Good for basic needs | Free+ | | **whisper-large-v3-turbo** | High accuracy, fast processing | Basic+ | **Use Cases:** Meeting transcription, podcast processing, voice commands, accessibility ### 🎨 Image Generation (NEW CATEGORY) State-of-the-art diffusion models for visual content: | Model | Architecture | Plan Tier | |-------|--------------|-----------| | **sdxl** | Stable Diffusion XL | Professional+ | | **flux-schnell** | Ultra-fast Flux | Professional+ | | **sd-3.5-large** | Latest Stable Diffusion premium | Enterprise | **Use Cases:** Product mockups, marketing assets, concept art, creative prototyping ### 🔍 Embeddings (NEW CATEGORY) Vector embeddings for semantic search and RAG: | Model | Dimensions | Plan Tier | |-------|------------|-----------| | **all-minilm-l6-v2** | 384 | Free+ | | **bge-large-en** | 1024 | Basic+ | **Use Cases:** Semantic search, document similarity, recommendation engines, RAG pipelines ### 📝 General Text Enhanced text generation models: | Model | Size | Plan Tier | |-------|------|-----------| | **gemma-2b** | 2B | Free+ | | **qwen-7b** | 7B | Basic+ | | **gemma-9b** | 9B | Professional+ | **Use Cases:** Chat interfaces, content generation, summarization, Q&A ## Tier Distribution We've distributed these models across subscription tiers to ensure every developer has access: | Tier | Models Available | Total Count | |------|------------------|-------------| | **Free** | gemma-2b, whisper-small, all-minilm-l6-v2, NousCoder, LLAMA 3.3-8B | 5+ base models | | **Basic** | +qwen-7b, qwen-coder-7b, whisper-turbo, bge-large, LLAMA 3.3-70B | +5 models | | **Professional** | +gemma-9b, deepseek reasoning (2), qwen-coder-32b, sdxl, flux-schnell, Claude Sonnet | +6 models | | **Enterprise** | +deepseek-r1, sd-3.5-large, Claude Opus | +3 models | ## Same API, More Capabilities All models use the same Chat Completion endpoint: ```bash POST https://api.ainative.studio/v1/managed/chat/completions ``` ### Text/Chat Example ```json { "preferred_model": "qwen-coder-7b", "messages": [ {"role": "user", "content": "Write a function to calculate fibonacci"} ], "max_tokens": 500 } ``` ### Response Format ```json { "id": "chatcmpl-abc123", "model": "qwen-coder-7b", "provider": "huggingface", "created": 1736809200, "choices": [{ "index": 0, "message": { "role": "assistant", "content": "def fibonacci(n):\n if n <= 1:\n return n\n return fibonacci(n-1) + fibonacci(n-2)" }, "finish_reason": "stop" }], "usage": { "prompt_tokens": 15, "completion_tokens": 35, "total_tokens": 50 }, "credits_consumed": 0.08, "credits_remaining": 999.92, "plan_tier": "basic" } ``` ## Transparent Credit Pricing Every model has clear, usage-based pricing: | Category | Base Cost | Token Cost | Unit | |----------|-----------|------------|------| | **Text/Chat** | 0.04 - 0.1 | +0.01-0.05 per 1K | credits | | **Code** | 0.05 - 0.12 | +0.01-0.06 per 1K | credits | | **Reasoning** | 0.08 - 0.2 | +0.02-0.1 per 1K | credits | | **Audio** | 0.02 - 0.08 | per minute | credits | | **Image** | 0.5 - 2.0 | per image | credits | | **Embeddings** | 0.01 - 0.03 | per 1K tokens | credits | Your credit balance and consumption is included in every API response - no surprises. ## Start Building Today All 15 models are live in production. Log in to your [AINative Studio dashboard](https://www.ainative.studio/dashboard) to explore: - **DeepSeek R1** for complex reasoning - **Qwen Coder** for specialized code generation (alongside NousCoder) - **Whisper** for audio transcription - **SDXL** for image creation - **BGE** for semantic search Each model is production-ready, fully tested, and backed by reliable serverless infrastructure. ## What's Next This expansion is just the beginning. We're constantly evaluating new models and capabilities to add to the platform. Need a specific model? [Open an issue on GitHub](https://github.com/AINative-Studio/core/issues). Ready to build multimodal AI applications? [Get started with AINative Studio](https://www.ainative.studio/signup). --- **About This Update** All models are accessible through our managed Chat Completion API with zero API key management required. Start building multimodal AI applications in minutes with our unified API interface.

Chat Completion API Expansion: 15 New Models Across 6 Modalities

Check your site's AX Score