Z.ai's top-tier GLM-5.3 Prime: 1M-token context, built for long-horizon coding and agent work. Served via OpenRouter.

Inception's Mercury 3: diffusion-based language model built for low-latency agentic workflows, 1M-token context. Served via OpenRouter.

MiniMax M3.1 Flash: fast multimodal model with 1M-token context, strong at agentic tasks and coding. Served via OpenRouter.

Moonshot AI's Kimi K3.1 Pro: open-weight multimodal reasoning model for long-horizon coding, 1M-token context. Served via OpenRouter.

OpenAI's GPT-6.1 Terra Pro: GPT-6.1 Terra served in pro reasoning mode for harder tasks, 1.05M-token context. Served via OpenRouter.

OpenAI's GPT-6.1 Terra: the mid-tier model of the GPT-6.1 series with 1.05M-token context. Served via OpenRouter.

StepFun's flagship model for agentic work: sparse Mixture-of-Experts (27B active / 600B total), strong at software engineering and professional tasks, 1M-token…

Cohere's Command A+: enterprise model for agentic, RAG and multilingual workloads with 192K-token context. Served via OpenRouter.

Meta's Muse Spark Open: open-weights Muse Spark variant with 1M-token context. Served via OpenRouter.

Meta's Muse Wonder: large Muse reasoning model with 1M-token context. Served via OpenRouter.

OpenAI's GPT-6.1 Sol Pro: the Pro tier of GPT-6.1 Sol with 1.05M-token context. Served via OpenRouter.

Google's Gemini 4 Argon: reasoning model with 1M-token context and multimodal input. Served via OpenRouter.

Inkling — frontier text model from Thinking Machines Lab with a 524k context window.

DeepSeek V4.1 Flash — fast, low-cost 1M-context text model from DeepSeek.

MiniMax M3 — long-context (1M) frontier text model from MiniMax.

Mistral's new flagship model with a 1M-token context window.

GPT-6.1 Sol from OpenAI — a flagship general-purpose model with a 1M-token context window.

Claude Haiku 5.5 from Anthropic — the fastest and most affordable Claude, built for high-volume everyday tasks.

Claude Sonnet 5.5 from Anthropic — a fast, capable all-round model for writing, coding and analysis with a 1M-token context window.

Qwen3.8 Omni Flash is Alibaba's first Qwen model built around agentic capabilities with native omni-modal audio-video understanding, 1M context, and tool use.

GPT-6 Luna is OpenAI's budget tier in the GPT-6 family, recently price-cut to $0.10/$0.50 per million tokens.

OpenAI's GPT-6 mid tier — successor to GPT-5.6 Sol, at OpenAI's current post-price-cut rate.

OpenAI's GPT-6 flagship model, a new generation above the GPT-5.6 family.

xAI's current flagship coding and knowledge model, GA September 2026, with text + image input and a 500K token context window.

Anthropic's most capable widely released model — successor to Claude Fable 5, for the most demanding reasoning and long-horizon agentic work, with a 1M token c…

Anthropic's Opus 5.5 — beats Fable 5.1 on agentic benchmarks at 60% lower API price, with a 1M token context window by default.

Meta's Muse Spark 1.3: multimodal reasoning model tuned for longer-horizon agentic, multi-agent and coding workflows, ~20% fewer tool calls than 1.2. Omni-moda…

Google's most intelligent Flash-tier model, GA 2026-09-02: significant gains over 3.7 Flash on software engineering, agentic tasks and multi-step reasoning. 1.…

Mistral AI's current flagship: 675B/41B-active MoE, Apache 2.0 open weights, 262K context, tool calling and structured outputs.

Meta's Muse Spark 1.2 flagship: omni-modal input (text, image, video, audio, files), 1M-token context, Meta's successor to the Llama line. Served via OpenRoute…

Tencent's Hunyuan Hy4 (preview): 1M-token context text model with switchable reasoning, the newest Hunyuan LLM tier. Served via OpenRouter.

Alibaba's fast, low-cost Qwen3.8 tier: multimodal input, 1M-token context, tuned for coding, agents and vision tasks. Served via OpenRouter.

Alibaba's Qwen3.8 Max flagship: natively multimodal (text, image, video input), 1M-token context, deep reasoning. Served via OpenRouter.

Z.ai's flagship GLM-5.3: hybrid sparse + linear attention, 1.3M-token context, strongest GLM tier for long-horizon coding and agent work. Served via OpenRouter.

Zhipu AI's 320B-A18B natively multimodal MoE with a 1M token context window, supersedes the GLM-5.2 base tier.

Google's coding/agent-focused Flash tier with a 1.05M token context window, GA since 2026-08-13.

Z.AI's open-weights flagship: beats GPT-5.5 on multiple long-horizon coding benchmarks, 1M token context, 131K output.

Meta Muse Glimmer 30B is an Apache-2.0 open-weight 30B dense agentic model from Meta, built for long tool-call sequences, hosted on Together AI.

Gemini 3.1 Pro is Google's current Pro-tier flagship reasoning model, GA on the Gemini API with a 1M token context window.

GPT-5.6 Luna is OpenAI's budget tier in the GPT-5.6 family, offering a 1.05M token context window at a lower price point than Sol or Terra.

OpenAI GPT-5.6 Terra — balanced mid-tier reasoning model between Sol (flagship) and Luna (cost-efficient)

Moonshot AI's current flagship: 2.8T-parameter MoE (16-of-896 experts active), Kimi Delta Attention, native vision, 1M token context.

Google's fastest, cheapest current Flash tier (~350 tok/s) — GA July 2026, for high-volume low-latency text tasks.

Google's current Flash-tier model, GA July 2026 — successor to Gemini 3.5 Flash with faster throughput and the same agentic/coding focus.

Anthropic's newest Opus-tier flagship, released July 2026 — successor to Opus 4.8 with the same pricing tier, built for long-horizon agentic work, knowledge ta…

Moonshot AI's 1T-parameter MoE flagship (32B active), 256K context

xAI's current flagship model, trained with Cursor

Alibaba's flagship closed-weight Qwen text model with a 1M-token context window, agent-grade reasoning and tool use.

DeepSeek's fast, low-cost flagship model with 1M token context and dual thinking/non-thinking modes

Anthropic's Claude Sonnet 5 — near-Opus quality on coding and agentic work at Sonnet-tier speed and cost, with adaptive thinking.

OpenAI's GPT-5.6 flagship reasoning model

xAI's reasoning flagship model, successor to Grok 4

DeepSeek's flagship reasoning model with 1M token context

Google's most intelligent Flash-tier model for sustained frontier performance on agentic and coding tasks. Fully GA as of May 2026, succeeding Gemini 2.5 Flash.

Anthropic's fastest and most cost-effective current model — ideal for simple, speed-critical tasks. The platform's first Anthropic fast/cheap tier.

Anthropic's most capable widely released model — for the most demanding reasoning and long-horizon agentic work. A new tier above Opus, with a 1M token context…

Anthropic's most capable Opus-tier model — highly autonomous, state-of-the-art on long-horizon agentic work, knowledge work, and memory, with clearer, warmer w…

OpenAI GPT-5.4 — latest generation language model with improved reasoning and creativity

Latest balanced AI model combining speed with intelligence—analyzes images, processes large documents, and excels at programming tasks with multilingual support

Latest balanced AI model combining speed with intelligence—analyzes images, processes large documents, and excels at programming tasks with multilingual support

Latest balanced AI model combining speed with intelligence—analyzes images, processes large documents, and excels at programming tasks with multilingual support

GPT-OSS 120B is OpenAI's open-weight 120-billion parameter language model designed for customization, on-premise deployment, and full enterprise control. This …

Llama 3.3 70B Instruct Turbo by Meta is a powerful open-source instruction-tuned AI language model with 70 billion parameters, optimized for extended context u…

Fast factual search & reasoning; also multilingual multimodal embeddings

Real-time web search and synthesis, fast, cited answers

Efficient multimodal AI understanding images, audio, and video at high speed—cost-effective solution for content moderation, media analysis, and automated tran…

Google's most capable AI processing massive multimodal datasets with advanced reasoning—ideal for academic research, complex data analysis, and scientific comp…

Fast multimodal reasoning AI understanding images and text—combines visual analysis with mathematical logic for data science, engineering diagrams, and technic…

OpenAI's reasoning powerhouse with multi-step thought processes—solves advanced mathematics, writes complex algorithms, and performs deep scientific analysis

Compact reasoning model optimized for STEM tasks—delivers accurate solutions in mathematics, physics, chemistry, and programming at lower cost

Latest balanced AI model combining speed with intelligence—analyzes images, processes large documents, and excels at programming tasks with multilingual support

Latest balanced AI model combining speed with intelligence—analyzes images, processes large documents, and excels at programming tasks with multilingual support

OpenAI GPT-5.4 — latest generation language model with improved reasoning and creativity

Latest iteration with breakthrough reasoning abilities and multimodal understanding—excels at complex problem-solving, advanced mathematics, and enterprise-lev…

Efficient multimodal AI processing images and large documents at high speed—optimized for rapid content generation, summarization, and real-time applications

GPT-5 represents OpenAI's next-generation flagship AI model with breakthrough capabilities in advanced reasoning, multimodal understanding, and sophisticated c…

Advanced multimodal AI understanding text, images, video, and large files with superior coding capabilities—excellent for technical documentation, data analysi…

Generates text, understands images and code; excels at reasoning

GPT-4o (GPT-4 Omni) by OpenAI is a lightning-fast, cost-efficient multimodal AI model that processes both text and images with exceptional contextual understan…

GPT-4o Mini is OpenAI's most cost-effective multimodal AI model, offering an optimal balance between performance and affordability. This compact yet powerful m…