Models Catalog
Explore this guide to learn how ChatboxAI works and how to integrate it into your workflow.
Models Catalog
ChatboxAI provides a unified, production-ready model catalog under the chatboxai/ namespace. By authenticating with your cbx_live_* key and passing a chatboxai/ model ID, you can instantly switch between the world's best models without altering your application code.
1. Normal Search & Chat Models
These models are optimized for everyday conversational AI, answering questions, writing text, extracting data, and processing normal queries.
Free Tier (Normal Chat)
These models are available for free (limited to 300 requests per day) and are great for high-speed, cost-effective tasks.
| Model Name | Model ID | Capabilities & Purpose |
|---|---|---|
| Llama 3.3 70B | chatboxai/llama-3.3-70b | Purpose: Best overall open-source model. Excellent for general chat, creative writing, and robust instruction following. |
| Gemini 2.5 Flash | chatboxai/gemini-2.5-flash | Purpose: Extremely fast, next-generation multimodal model. Best for high-speed text and vision tasks. |
| Qwen3 32B | chatboxai/qwen3-32b | Purpose: Strong coding and logic reasoning. Ideal for general-purpose programming queries. |
| Qwen 3.6 Plus Preview | chatboxai/qwen-3.6-plus | Purpose: Advanced multi-turn conversation and coding logic. |
| Mistral Small 3.1 | chatboxai/mistral-small-3.1 | Purpose: Fast and lightweight with native function calling support. |
| GLM 4.5 Air | chatboxai/glm-4.5-air | Purpose: Built-in chain-of-thought processing for structured problem solving. |
| Llama 3.1 8B Turbo | chatboxai/llama-3.1-8b | Purpose: Ultra-fast, high-throughput model. Best for simple queries and API fallback. |
| Gemma 3 27B | chatboxai/gemma-3-27b | Purpose: Highly capable, balanced text and vision processing. |
(Note: Additional lightweight models like llama-3.2-3b, qwen3-next-80b, minimax-m2.5, and nemotron-3-nano are also included in the free tier for edge-case tasks).
Paid Tier (Normal Chat)
These are flagship models offering state-of-the-art intelligence, massive context windows, and high reliability.
| Model Name | Model ID | Capabilities & Purpose |
|---|---|---|
| Claude Opus 4.7 | chatboxai/claude-opus-4.7 | Purpose: Flagship reasoning. Best for highly complex problem solving, deep analysis, and high-fidelity content generation. |
| Claude 4.5 Sonnet | chatboxai/claude-4.5-sonnet | Purpose: Perfect balance of speed and top-tier intelligence. Ideal for versatile everyday tasks. |
| Gemini 2.5 Pro | chatboxai/gemini-2.5-pro | Purpose: Premium performance with massive context (up to 2M tokens). Best for analyzing entire codebases or huge document sets. |
| Qwen3 Coder 480B | chatboxai/qwen3-coder | Purpose: Heavy-duty coding. Best for building complex applications and debugging huge codebases. |
| GPT-OSS 120B | chatboxai/gpt-oss-120b | Purpose: Top-tier open-source synthesis and architectural design. |
2. Deep Research & Agentic Models
These models are explicitly built for multi-step reasoning, autonomous agents, and deep internet research workflows. They excel at using external tools, browsing the web, and synthesizing long-form research.
Free Tier (Deep Research)
| Model Name | Model ID | Capabilities & Purpose |
|---|---|---|
| Kimi K2 | chatboxai/kimi-k2 | Purpose: Highly optimized for agentic workflows and tool usage. Capable of handling deep web searches and synthesizing information autonomously. |
Paid Tier (Deep Research)
| Model Name | Model ID | Capabilities & Purpose |
|---|---|---|
| NVIDIA Nemotron 3 Super | chatboxai/nemotron-3-super | Purpose: Massive 1M context window combined with deep agentic reasoning. Best for orchestrating complex research workflows over massive document libraries. |
| Hermes 3 405B | chatboxai/hermes-3-405b | Purpose: Frontier reasoning and strict tool-use. Best for handling state-of-the-art multi-step logic puzzles and autonomous agents. |
| Trinity Large Preview | chatboxai/trinity-large | Purpose: Massive 400B MoE model. Best for expansive, complex persona adoption during agentic interactions. |
3. Image Generation Models
Used exclusively with the POST /api/v1/images/generations endpoint to generate high-quality visuals from text prompts.
Free Tier (Image Generation)
| Model Name | Model ID | Capabilities & Purpose |
|---|---|---|
| FLUX Schnell | chatboxai/image-flux-schnell | Purpose: High-speed image generation. Ideal for rapid visual prototyping and quick concept art. |
Paid Tier (Image Generation)
| Model Name | Model ID | Capabilities & Purpose |
|---|---|---|
| FLUX Dev | chatboxai/image-flux-dev | Purpose: Ultra-high-quality image generation. Best for production-ready artwork, hyper-realistic photos, and detailed visuals. |
| Stable Diffusion 3.5 | chatboxai/image-stable-diffusion-3.5 | Purpose: Highly flexible generation. Best for complex prompt adherence and highly stylized artistic rendering. |
| Stable Diffusion XL | chatboxai/image-sdxl | Purpose: Cinematic and photographic visuals. Best for highly realistic character portraits and environments. |
4. Audio & Speech Models (Bonus)
Used exclusively with the POST /api/v1/audio/speech endpoint for Text-to-Speech (TTS).
| Tier | Model Name | Model ID | Capabilities & Purpose |
|---|---|---|---|
| Free | TTS-1 | tts-1 | Purpose: Standard text-to-speech for everyday voice generation and basic app accessibility features. |
| Paid | TTS-1 HD | tts-1-hd | Purpose: High-definition, low-noise audio generation. Best for professional voiceovers, podcasts, and media production. |
Auto Selection
Don't want to choose? Pass "model": "auto" to let ChatboxAI pick the best available model for your request dynamically based on latency and availability.
{ "model": "auto", "messages": [...] }Model ID Stability
chatboxai/ model IDs are strictly maintained for production reliability. We guarantee:
- The model ID string will never randomly change.
- The behavior and quality tier remain consistent.
- If we upgrade infrastructure internally, your application code requires zero updates.
See Pricing for detailed cost information for each model tier.