TokenRouter @TokenRouter_US

GLM 5.2 (full 1M context) is 60% off on TokenRouter through July 19. Stack our Top Up 7, Get 3 bonus and pay as little as 28% of list price. Learn more at tokenrouter.com/models

GPT‑5.6 is available now on TokenRouter! Experience Sol, Terra, and Luna now through our unified API endpoint. Whether you need Sol's frontier-level performance for your toughest coding, research, or agentic workflows, Terra's balanced everyday capability, or Luna's speed and affordability for high-volume tasks, GPT‑5.6 gives you a model for every use case and budget, all accessible through TokenRouter's unified API.

Grok 4.5 is available now on TokenRouter. SpaceXAI's smartest model built for coding, agentic tasks, and knowledge work.

Fable 5 is back on TokenRouter!

TokenRouter is now a supported LLM backend in @IBM's AssetOpsBench, integrated into its out-of-the-box agents and MCP server. AssetOpsBench is IBM's open framework for Industry 4.0 asset operations, with specialist agents covering IoT, time-series forecasting, FMSR, and work order management, plus multi-agent orchestration over MCP. Teams running the benchmark can now point those agents at a single TokenRouter endpoint and reach 100+ models across OpenAI, Anthropic, Google, DeepSeek, MiniMax, Qwen, and others. For industrial AI teams figuring out which models actually hold up on asset operations work, that takes out the rewrite step every time you want to put a new provider through the benchmark. Thanks to Dhaval Patel and the AssetOpsBench team at @IBMDeveloper, including Shuxin Lin, Miaomiao Feng, and Chathurangi Shyalika, for adding TokenRouter support. Interested in AssetOpsBench? Star the GitHub repo to support the project, and fork it to start experimenting: lnkd.in/g2M9eiy6

👀 Fable 5 tomorrow?

Sonnet 5 is now live on TokenRouter! It is the most agentic model yet in the sonnet-class models, with significant improvement from over its predecessor, sonnet 4.6 on important aspects of agentic performance like reasoning, tool use, coding, and knowledge work, according to @AnthropicAI. Try it now on tokenrouter.com Seasonal 30% off applies platform-wide, all models, no markup added on our end. Terms apply.

What a day at Agent Forge Hackathon at @sgSMU during @superai_conf Week! It was amazing to see teams use TokenRouter to route between top AI models, leverage smart routing based on tasks, and ship real working agents in just 5 hours. These are the builders shaping the future of AI, and we're glad our platform could be part of their stack. 🚀 Many thanks to @devonsun_ml at @theaibuilders, SMU Artificial Intelligence Club (SMUAI), and every participant who built with us. See you at the next one. #AgentForge #TokenRouter #AI #Hackathon #AIAgents #Singapore #AIBuilders #SMUAI

keyboard_arrow_left Previous keyboard_arrow_right Next

Singapore is quickly becoming one of Asia's most dynamic AI hubs, powered by a thriving ecosystem of researchers, entrepreneurs, startups, and investors. And we saw it firsthand at @superai_conf Week. Grateful to have joined a conversation around bridging Silicon Valley and Singapore during @EpicConnector 's Demo Night & Afterparty. Our CEO, Stephen Watts, spoke on the panel about empowering Singapore entrepreneurs with efficient and stable infrastructure, an important conversation as AI founders continue building and scaling across Singapore and beyond. Thank you to the organizers, fellow speakers Kisson Lin, Ang Li, Christoph Burgdorfer, founders, investors, and everyone we connected with throughout the evening. Excited to keep supporting the next generation of AI builders. #SuperAI #SingaporeAIWeek #SingaporeStartups #AIInfrastructure #Entrepreneurship #AI #TokenRouter

keyboard_arrow_left Previous keyboard_arrow_right Next

We heard you, so @MiniMax_AI Days is getting extended. M3 will stay free on TokenRouter through June 22. And every other MiniMax model is still 50% off. Explore on TokenRouter: tokenrouter.com

Z.ai GLM-5.2 is here, a flagship open model designed for long-horizon tasks, with a 1M-token context window, and stronger coding capabilities. You can now access GLM-5.2 through TokenRouter’s unified API and start testing it alongside other leading models. Explore on TokenRouter: tokenrouter.com

Following the recent suspension of Claude Fable 5 access, TokenRouter will be taking down Claude Fable 5 from our platform. What this means for users: Existing requests to Claude Fable 5 may return errors, and we recommend updating your integrations to another available Claude model as soon as possible. All other Claude models on TokenRouter remain available. Our team is monitoring the situation closely and will continue to keep model availability updated in real time. Thank you for your understanding.

@hiut1u @MiniMax_AI None.

@MiniMax_AI M3 is free on TokenRouter through June 17. Every other MiniMax model is 50% off. Visit at tokenrouter.com

@badaboo72603773 @MiniMax_AI This is not our website. Please visit tokenrouter.com

@vidhanmertiya @MiniMax_AI Hi this is a great question! The label reflects output modality — M3 is a text-output model (it can take image inputs/understand images, but generates text responses, not images). So ‘Text’ here means what it produces, not what it can read. Hope that clears it up!

@imshreekumar @MiniMax_AI Hi, is not our page, please visit tokenrouter.com

@mikascend updated！

Get your free developer welcome credit at tokenrouter.com. Unlock access to every frontier models under one unified billing. Authentic models ☑️ Cost transparency ☑️ Smart routing ☑️ Switch your base url today.

@TommyBez85 @aisdk Yes. TokenRouter exposes an OpenAI-compatible endpoint, so it works with the AI SDK through the openai-compatible provider. Just point the baseURL at TokenRouter and pass your API key.

@static_echo_io @aidigipt @MiniMax_AI Should be resolved now, thanks.

@aidigipt @MiniMax_AI Should be resolved now!