All Models
Browse all available AI models — compare pricing, features, and start integrating.
Xiaomi MiMo-V2.6-Pro activates 42B parameters per token for complex engineering tasks. Call as mimo-v2.6-pro.
Xiaomi MiMo-V2.6-Flash delivers 1M context at ~1/3 the cost of Pro. Call as mimo-v2.6-flash.
Uncompromised Muse Spark 1.3 checkpoint with 1M context, vision, and tool calling at 90%+ off standard rates. Call as muse-spark-1.3-contributor.
Use MiniMax M3 when documents or agent memory need to stay in one request.
Use GLM 5.2 when reasoning trace quality and structured conclusions matter.
Use Doubao Seed 2.1 Turbo for fast Chinese-language production traffic.
Use Kimi K2.7 Code when code context is the bottleneck.
Use DeepSeek V4 Flash as the first-pass route for cost-sensitive technical traffic.
Use DeepSeek V4 Pro when a prompt needs deeper reasoning and a careful final recommendation.
Calibrated yes/no confidence (noul), choice selection, and score interpolation at $0.042/1M input tokens.
Claude Opus 5 is suited for demanding reasoning, advanced coding, analysis, long-form writing, and agent workflows through the Anthropic Messages API.
Gemini 3.6 Flash delivers fast, cost-efficient text generation for everyday chat, writing, summarization, and analysis.
Kimi K3 is designed for long-context reasoning, code understanding, writing, analysis, and agent-style workflows through an OpenAI-compatible Chat Completions API.
An AI model for complex reasoning, coding, and long-document analysis.
GPT-5.6 Sol is a premium text model for demanding coding, reasoning, and long-form agent work.
GPT-6.1 Sol is OpenAI's upgrade to GPT-6 Sol for demanding work across code, documents, and applications. It keeps the 1M-token context window and 128K output, supports reasoning effort up to max, and halves the price of cached input.
GPT-5.6 Terra is a stronger text model for reasoning-heavy coding and analysis tasks.
GPT-5.6 Luna is a balanced text model for everyday coding, writing, and agent workflows.
Claude Sonnet 5 balances intelligence, speed, and cost for advanced reasoning, coding, writing, and everyday professional work.
Claude Fable 5 is a premium model for deep reasoning, complex analysis, long-form creation, and demanding agent workflows.
Claude Opus 4.8 is designed for demanding reasoning, coding, analysis, and professional knowledge workflows.
Multimodal intelligence for understanding diverse information.
DeepSeek V4.1 Flash is a new 552B MoE with 8B to 16B active parameters, 1M context, native image input, and the top Terminal-Bench and DeepSWE scores in the DeepSeek line. AIReiter bills it below the official off-peak rate with no peak-hour surcharge.
Qwen3.8 Max is a 1M-context flagship. Use Chat Completions for dialogue and PDF; use Responses for web_search, code_interpreter, and image search.