Fast OpenRouter-hosted GLM 5 Turbo variant from Z-AI for general chat, coding, and tool use. Not included in the subscription.

Qwen3.5 27B is a native vision-language dense model optimized for fast responses while balancing quality and inference speed. Running inside a TEE (Trusted Execution Environment), with verifiably no logging by the provider.

Grok 4.20 Beta with reasoning enabled. xAI positions it as a flagship model with fast agentic tool calling, strong prompt adherence, and a 2M-token context window.

Grok 4.20 Beta with reasoning disabled for lower-latency tool-centric workloads, while keeping the same 2M-token context window and multimodal input support.

Grok 4.20 Multi-Agent Beta tuned for agentic workflows with tool calling, structured outputs, reasoning support, and a 2M-token context window.

Nvidia's Nemotron 3 Super 120B A12B model from the March 2026 Nemotron 3 release. It uses a hybrid Mamba-Transformer MoE architecture and targets agentic and coding workloads with a 262K context window here.

Nvidia Nemotron 3 Super 120B with reasoning content enabled. Returns separate thinking content when requested.

ByteDance Seed 2.0 Lite is a balanced long-context model for high-frequency enterprise workloads, tuned for unstructured information processing, text creation, search and recommendation, and stable structured outputs. Supports 262k context on OpenRouter and routes via BytePlus Singapore, which claims no logging.

GPT-5.4 is OpenAI's latest frontier model for professional work with stronger reasoning, coding, and tool use.

The highest-performing GPT-5.4 variant from OpenAI, tuned for complex reasoning and demanding professional tasks.

Chat-optimized GPT-5.3 variant with improved conversational quality and instruction following.

Google's Gemini 3.1 Flash Lite Preview via OpenRouter. Optimized for high-volume, cost-efficient reasoning and multimodal workloads.

Upstage's Solar Pro 3 is a Mixture-of-Experts (MoE) language model with 102B total parameters and 12B active parameters per forward pass, optimized for Korean with strong English and Japanese support.

Qwen 3.5's open-source 397B MoE model (17B active params) with hybrid linear attention. Running inside a TEE (Trusted Execution Environment), with verifiably no logging by the provider.

Gemini 3.1 Pro preview variant tuned for better tool selection behavior in coding agents and multi-tool workflows. It reduces overuse of generic bash tools and improves function-calling reliability while retaining Gemini 3.1 Pro's multimodal reasoning and 1M-token context. NOTE: Inputs > 200k tokens are charged at 2x input and 1.5x output rates.

A 24B-parameter Mixture-of-Experts model from LiquidAI with 2B active parameters per token, designed for efficient high-quality generation and practical local deployment.

Coding-focused GPT-5.3 variant with optimized routing.

Qwen3.5 122B A10B is Alibaba's high-end native vision-language model in the Qwen 3.5 family with strong text and multimodal performance.

Qwen3.5 122B A10B with extended reasoning enabled. Alibaba's high-end native vision-language model in the Qwen 3.5 family.

Qwen3.5 27B is a native vision-language dense model optimized for fast responses while balancing quality and inference speed.

Qwen3.5 27B with extended reasoning enabled. A native vision-language dense model optimized for fast responses.

Qwen3.5 35B A3B is a native vision-language MoE model with hybrid attention designed for efficient inference and strong general performance.

Qwen3.5 35B A3B with extended reasoning enabled. A native vision-language MoE model with hybrid attention.

Qwen3.5 Flash is the fastest and cheapest native Qwen 3.5 vision-language model with a 1M-token context window.

Qwen3.5 Flash with extended reasoning enabled. The fastest and cheapest native Qwen 3.5 vision-language model.

Aion-2.0 is a DeepSeek V3.2 variant tuned for immersive roleplay and storytelling with stronger tension, conflict, and darker thematic nuance.

Gemini 3.1 Pro preview high-reasoning variant.

Gemini 3.1 Pro preview low-reasoning variant.

Gemini 3.1 Pro preview is built for tasks where simple answers are not enough. Stronger core reasoning for complex coding, math, and long-context workflows, with multimodal support and a reported 77.1% verified score on ARC-AGI-2. NOTE: Inputs > 200k tokens are charged at 2x input and 1.5x output rates.

Claude Sonnet 4.6 is Anthropic's most capable Sonnet yet — a full upgrade across coding, computer use, long-context reasoning, agent planning, and design. Supports up to a 1M-token context window at Sonnet pricing.