Switch between GPT-5.6 Sol, Claude Opus 5, Gemini 3.6 Flash, Llama 4, and 200+ more — your memory follows seamlessly.
// Monday: Use GPT-5.6 Sol
model: "gpt-5.6-sol"
// Tuesday: Try Claude
model: "anthropic/claude-opus-5"
// Wednesday: Test Gemini
model: "google/gemini-3.6-flash"
// Memory persists across all of them ✓No lock-in. No migration. Just change the model parameter.
OpenAI flagship for complex professional reasoning, coding, research, and computer use. 128K max output.
Balanced GPT-5.6 model for everyday production work, with frontier tools and lower cost than Sol.
Fast, cost-efficient GPT-5.6 model for high-volume workflows and multi-step tool use.
OpenAI voice model for natural, continuous conversation with frontier-model delegation for deeper work.
Anthropic premium model for serious coding, long-running agents, and complex professional work.
Anthropic's most agentic Sonnet: strong reasoning, browser and terminal use, coding, and knowledge work.
Fast, economical Claude model for responsive assistants and simple tasks at scale.
Google's stable agentic workhorse for coding, multimodal tasks, spatial reasoning, and rapid tool loops.
Frontier-speed model for subagents, multi-step workflows, and long-horizon coding tasks.
Low-latency, cost-efficient multimodal model for document parsing and high-throughput subagents.
Google Pro-class reasoning model for advanced knowledge work and multimodal workflows.
128B open-weight multimodal flagship for long-horizon coding, reasoning, and productivity agents.
Apache 2.0 hybrid model unifying instruct, reasoning, multimodal, and agentic coding capabilities.
Document-intelligence model with bounding boxes, confidence scores, and support for 170 languages.
General Grok model for fast chat, reasoning, and tool-heavy agents.
High-capability Grok 4 line model available through OpenRouter.
xAI build-focused model for implementation-heavy agent workflows.
1.6T-total, 49B-active open model for agentic coding, reasoning, and rich world knowledge.
Public-beta V4 Flash update with thinking modes, native Responses API support, and Codex optimization.
284B-total, 13B-active high-throughput model with thinking and non-thinking modes.
Perplexity search model for grounded answers and web-aware agents.
Perplexity reasoning model for grounded, multi-step research tasks.
Deep research model for long-form investigations and synthesis.
Cohere, Perplexity, Together, and dozens more. If it's on OpenRouter, it works with MemoryRouter.
One line change. Same code. Persistent memory across every model.
Start free with a 14-day trial. No credit card required.