TokenReply was originally called resurge.one. Later, since the .com and .ai domains were not available, we decided to rename the project to TokenReply.

TokenReply Update

Hi everyone,

I wanted to share another TokenReply update, since a lot has changed over the past few months.

Platform Status

TokenReply has now been running stably for over 3 months, and we have passed roughly 4500 registered users. The feedback from genuine users has honestly been much better than I expected. I am really grateful to everyone who stayed around while the project gradually became more stable, more usable, and easier to rely on.

At the same time, I also want to be transparent about the less fun side of running the platform. Out of those roughly 4500 registrations, around 600 turned out to be abusive or multi account signups. Most of them were created to farm free quota, automate usage, or resell access. Because of that, I have been actively working on cleanup, abuse detection, and better account protection systems. If the free tier has felt a bit tighter recently, this is the main reason. I am trying to make sure the budget goes toward real users instead of being drained by automated abuse.

Recently Added Free Models

• grok-4.3
• grok-4.20-0309-reasoning
• grok-4.20-0309-non-reasoning
• grok-4.20-multi-agent-0309
• deepseek/deepseek-v4-flash
• z-ai/glm-4.5-air
• minimax/minimax-m2.5
• meta-llama/llama-3.3-70b-instruct
• meta-llama/llama-3.2-3b-instruct
• google/gemma-4-31b-it
• google/gemma-4-26b-a4b-it
• nousresearch/hermes-3-llama-3.1-405b

These are available on top of the existing free model pool, which already includes GLM, DeepSeek, older Grok variants, and other rotating models.

Weekly Featured System

We also introduced a Weekly Featured system. Every user gets 40 conversations per day with featured models, while daily check in increases that limit to 50 conversations per day. Check ins also randomly grant between $0.00 and $0.20 worth of allowance.

The goal is to let people keep experimenting with new models without immediately needing to spend money, while still keeping the platform sustainable in the long term.

About the $5 Subscription

A few people have asked why the subscription costs $5 per month while also including $5 worth of balance. The idea is actually simple.

The subscription is designed to give users flexible access to all supported models without forcing complicated tier restrictions or locking specific models behind separate plans. With the subscription, you can freely choose whichever supported models you personally want to use, including Claude, GPT, Grok, DeepSeek, Kimi, GLM, Gemini, and others.

Many models are available through optimized routing at effectively 50% to 75% of their normal pricing. In some cases, certain routes can reach costs as low as around 20% of standard pricing, depending on provider availability and routing conditions.

The goal is to give users a simple and flexible way to explore different models while still keeping TokenReply sustainable over time. And honestly, if you only use free models, you genuinely do not need to subscribe. That is intentional.

The paid tier mainly exists for users who want access to premium models, better routing options, and lower effective pricing across different providers, without having to deal with complicated per token billing systems.

What I Still Care About

I still care a lot about keeping free models actually free whenever possible. As long as a provider continues offering a model for free, I will keep trying to make it available to everyone. If a provider removes free access, I will either remove the model or look for another stable upstream provider.

Still Actively Working On

• Better mobile UI
• Easier API key setup flow
• One click JanitorAI and SillyTavern imports
• Better bug report and feedback systems
• More stable and cheaper upstream providers
• Improved uptime and routing stability
• Better abuse and multi account detection, so legitimate users can get more headroom

How to Use TokenReply

Register

Create an account here:

https://www.tokenreply.com/

Create an API Key

After logging in, go to API Key Management and create an API key.

If you are not sure how to configure it, just fill in the Name field and click Submit.

Your API key should look like this:

sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Please keep your API key private and do not share it with others.

Use TokenReply in SillyTavern

In SillyTavern, set Base URL to:

https://api.tokenreply.com/v1

Set API secret key to your generated key:

sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Use TokenReply in JanitorAI

In JanitorAI, set Base URL to:

https://api.tokenreply.com/v1/chat/completions

Set API secret key to:

sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

How to Find Model Names

You can find the available models and their exact model names on the pricing page:

https://www.tokenreply.com/models
https://www.tokenreply.com/pricing

When setting up TokenReply in SillyTavern, JanitorAI, or other tools, copy the model name exactly as shown on the pricing page.

For example, you can try models like:

• grok-4.3
• z-ai/glm-4.5-air

The pricing page is also the best place to check which models are currently free, paid, or available through optimized routing.

Links

Website:
https://www.tokenreply.com/

Model List and Pricing:
https://www.tokenreply.com/models
https://www.tokenreply.com/pricing

API Base URL:
https://api.tokenreply.com/v1

Discord:
https://discord.gg/guNZa3aM2q

Feel free to share feedback, model requests, suggestions, or complaints. I read basically everything people send me, even if replies sometimes take a while.