Before you start
You need two programs on the same computer: SillyTavern and FreeLLMAPI. FreeLLMAPI runs in Docker or as a desktop app. Install it with one command:
Install FreeLLMAPI (macOS or Linux)
curl -fsSL https://freellmapi.co/install.sh | bash- Open the dashboard at
http://localhost:3001. - Add free keys from two or more providers on the Keys page. Google AI Studio, Groq and Mistral each give a free key.
- Copy the unified key from the Keys page. SillyTavern uses only this key.
The desktop app listens on http://127.0.0.1:31415. If you use the desktop app, use that address in place of http://localhost:3001 on this page.
Step 1: Connect SillyTavern
Open SillyTavern and click the API Connections icon (the plug) in the top bar. Then set these fields:
| Field | Value |
|---|---|
| API | Chat Completion |
| Chat Completion Source | Custom (OpenAI-compatible) |
| Custom Endpoint (Base URL) | http://localhost:3001/v1 |
| Custom API Key | Your unified key, <unified-key> |
| Enter a Model ID | auto, or a model ID from the table below |
Click Connect. SillyTavern reads /v1/models and fills the Available Models list. Do not add /chat/completions to the URL. SillyTavern adds the path itself.
auto in Enter a Model ID to let the router select a model that works now.Click Test Message to send a short prompt. If you get an answer, the connection works.
Step 2: Choose models for long chats
A roleplay chat sends the character card, the world information and the chat history with each message. A long chat can send tens of thousands of tokens each time. Select models with a large context window.
You can use these model IDs in Enter a Model ID:
auto: follows your fallback chain on the dashboard.auto:smart: ranks enabled models by intelligence. Use it for story quality.auto:fast: ranks enabled models by measured speed. Use it for quick replies.- A fixed model ID: gives the same writing style every time, but stops when that model's free limit is used.
| Model | Free on | Context | Free limits (best offer) | Tools | Vision |
|---|---|---|---|---|---|
| GLM 5.3 | APINex, UnoRouter | 1M | 5 req/min 10M weighted tokens/day shared across APINex free routes; 4x token multiplier; 5 requests/minute/account | Yes | No |
| GLM 5.3 Flash | APINex, BlazeAPI, NVIDIA NIM +1 more | 1M | 40 req/min 40 RPM | Yes | Yes |
| Gemini 3.8 Flash | APINex, Google AI Studio | 1M | 5 req/min 10M weighted tokens/day shared across APINex free routes; 3x token multiplier; 5 requests/minute/account | Yes | Yes |
| Qwen3.8 Flash Next | AMD Radeon Cloud, BlazeAPI, ElectronHub | 262K | Recurring daily account allowance | Yes | No |
| Qwen3.8 Flash | ElectronHub | 1M | $0.25/week shared recurring credit | Yes | Yes |
| Deepseek V4.1 Flash | APINex | 1M | 5 req/min 10M weighted tokens/day shared across APINex free routes; 3x token multiplier; 5 requests/minute/account | No | No |
| Gemini 3.7 Flash | Google AI Studio | 1M | 10 req/min 20 req/day 250,000 tokens/min ~3M | Yes | Yes |
| GPT-5.6 Luna | ElectronHub | 1M | $0.25/week shared recurring credit | Yes | Yes |
| Dots3 Note Preview | Kilo Gateway, OpenRouter | 512K | 20 req/min 50 req/day 50 req/day account-wide (1,000 after a $10 purchase) | Yes | Yes |
| DeepSeek V4 Flash | AMD Radeon Cloud, BazaarLink, ElectronHub | 1M | 10 req/min 50 req/day Shared rate limit | Yes | No |
| DeepSeek V4 Flash 0731 | ElectronHub, Router9 | 1M | 10,000 credits/month (~$1), shared; account-anniversary reset | No | No |
| Qwen3.8 27B | BlazeAPI, ElectronHub, Gizmo +3 more | 262K | 30 req/min 250 req/day 70,000 tokens/min ~6M | Yes | Yes |
Showing 12 of 184 matching free models. See all models.
Live data from the FreeLLMAPI catalog (version 2026.11.19, checked 19 November 2026). Free limits change often. The catalog is checked every day.
FreeLLMAPI keeps a conversation on one model for 30 minutes. This keeps the character voice the same across messages. If a limit forces a change of model, the next model continues from the same chat history.
Step 3: Set the generation settings
Open the AI Response Configuration panel (the sliders icon). Three settings have the largest effect on free tiers:
| Setting | Recommended value | Why |
|---|---|---|
| Context Size (tokens) | 16,000 to 32,000 | A smaller context sends fewer tokens per message. Free tiers often limit tokens per minute. |
| Max Response Length (tokens) | 400 to 800 | Long replies use the output quota fast. Some reasoning models also use part of this budget for hidden thinking. |
| Streaming | On | You see the text as it arrives. FreeLLMAPI streams on every provider that it can reach. |
SillyTavern also has a Prompt Post-Processing option for the Custom source. Some providers reject two system messages in a row. If you get errors about message roles, select a Semi-strict or Strict option. These options merge messages of the same role. The options marked "no tools" also remove tool calls. The options marked "with tools" keep them.
Rate limits in long roleplay
Each free tier has its own limits. For example, a free Gemini Flash key allows few requests per day, and a Groq key allows more requests with lower token limits. One evening of roleplay can send several hundred requests. One free key does not last that long.
- Add more providers. Each key adds requests per day. FreeLLMAPI counts the quota of each key and moves to the next key or provider when one is used.
- Reduce the context. Summarize old messages with the SillyTavern Summarize extension. A shorter prompt uses fewer tokens per minute.
- Avoid swipes in a loop. Each swipe or regenerate is a new request.
- Watch the reset. When all keys are at their limit, the gateway returns error 429 with a
Retry-Afterheader. Many daily limits reset at midnight UTC.
The table shows the free limits per provider in today's catalog.
| Provider | Free models | Max req/min | Max req/day | Max tokens/min | Tokens/month (one model) | Strongest free model |
|---|---|---|---|---|---|---|
| ElectronHub | 84 | - | - | - | $0.25/week shared recurring credit | Qwen3.8 Flash (ElectronHub)qwen3.8-flash |
| Cloudflare Workers AI | 39 | - | - | - | ~5-10M | Qwen 3.8 27B (Cloudflare Workers AI)@cf/qwen/qwen3.8-27b |
| Inferbase | 38 | - | - | - | 5,000 requests/month; 100 requests and $0.50 usage/day shared | DeepSeek V4 Flash (fp8) (Inferbase)deepseek-v4-flash:fp8 |
| AI Horde | 31 | - | - | - | Ongoing volunteer queue; availability varies | Gemma 4 31B Heretic (AI Horde)koboldcpp/gemma-4-31B-it-heretic |
| ACLIDE | 29 | - | - | - | €20/month shared internal usage credits · no card | Qwen3.8 Max (ACLIDE)qwen/qwen3.8-max:free |
| Google AI Studio | 21 | 15 | 1,000 | 250,000 | ~3M | Gemini 3.8 Flashgemini-3.8-flash |
| Api.Airforce | 20 | 1 | 1,000 | - | free · 1,000 req/day · 1 req/min | gpt-oss-20b (Api.Airforce)gpt-oss-20b |
| NVIDIA NIM | 19 | 40 | - | - | free · 40 RPM | GLM 5.3 Flash (NV)z-ai/glm-5.3-flash |
| Cohere | 17 | 20 | 33 | - | ~1-2M | Command A+ (05-2026)command-a-plus-05-2026 |
| DreamPrompting | 14 | 100 | 5,000 | - | free · 500K tokens/day · 5,000 req/day | Gemini 3.1 Flash-Litegoogle/gemini-3.1-flash-lite |
| Mistral | 12 | 2 | - | 500,000 | ~50-100M | Mistral Codemistral-code-latest |
| ModelScope | 11 | - | 200 | - | Daily login: 200 credits + 50 for linked Alibaba Cloud; 24h expiry; shared; noncommercial experience | GLM-5.2 (ModelScope)ZhipuAI/GLM-5.2 |
Live data from the FreeLLMAPI catalog (version 2026.11.19, checked 19 November 2026). Free limits change often. The catalog is checked every day.
Troubleshooting
| Problem | Cause | Fix |
|---|---|---|
| Connect fails or the status stays red | Wrong URL or FreeLLMAPI is not running | Open http://localhost:3001 in the browser. Make sure that the URL ends in /v1. |
| Error 401 | The key is wrong | Paste the unified key again in Custom API Key |
| A selected model always fails | No key for that model's provider | Add the provider key, or type auto |
| Error 429 after a long session | All keys are at their limit | Add keys from more providers, or wait for the reset |
| Errors about message order or roles | The provider has strict role rules | Set Prompt Post-Processing to Semi-strict or Strict |
| Replies stop in the middle | Max Response Length is too small for a reasoning model | Increase the value, or select a model without reasoning |
| Character voice changes during the chat | A limit forced a change of model | Pin one model that is free on several providers |
Frequently asked questions
Is there a free API for SillyTavern?
Yes. SillyTavern accepts any OpenAI-compatible API through the Custom (OpenAI-compatible) source. FreeLLMAPI gives one endpoint for many free provider tiers. Each tier has its own limits.
What is the base URL for SillyTavern Custom (OpenAI-compatible)?
Use http://localhost:3001/v1 for a FreeLLMAPI server install, or http://127.0.0.1:31415/v1 for the desktop app. Do not add /chat/completions.
Which free model is best for SillyTavern roleplay?
Use a model with a context window of 128K tokens or more. Start with auto:smart for quality or auto:fast for speed. The table on this page shows the strongest free models today.
Why do I get error 429 in SillyTavern?
A free tier reached its limit. FreeLLMAPI moves to the next key or provider. If every key is at its limit, you get 429. Add more provider keys or reduce the context size.
Does SillyTavern streaming work with free APIs?
Yes. Turn on Streaming in the AI Response Configuration panel. FreeLLMAPI streams the reply from each provider in the OpenAI format.
Related guides
- Free LLM API for Open WebUI
- Free LLM API with 1 million token context
- Free LLM API rate limits
- Free OpenAI-compatible API base URLs
- Free LLM API with no credit card
Sources
- SillyTavern docs: Chat Completion sources, Custom (OpenAI-compatible) (checked 20 November 2026)
- SillyTavern source: API connection panel labels (public/index.html) (checked 20 November 2026)
- FreeLLMAPI docs: REST API (routing strategies, model listing, streaming) (checked 20 November 2026)
- FreeLLMAPI README: features and limitations (checked 20 November 2026)