Available models
Default - OpenRouter (Auto)
Default - OpenRouter (Auto)
Model ID:
openrouter/autoOpenRouter automatically selects the best available model for your request based on current availability, performance, and cost. This is the recommended option if you are unsure which model to use — it adapts without any input from you.After analysis completes, the UI shows the actual model that OpenRouter routed to, so you always know which model produced your verdict.OpenRouter (Free)
OpenRouter (Free)
Model ID:
openrouter/freeOpenRouter routes your request to a free-tier model from its catalog. The specific model may vary. Use this when you want free-tier access without specifying a provider.LiquidAI: LFM2.5-1.2B-Instruct (Free)
LiquidAI: LFM2.5-1.2B-Instruct (Free)
Model ID:
liquid/lfm-2.5-1.2b-instruct-20260120:freeProvider: LiquidA compact, free-tier instruction-following model from LiquidAI. Suitable for lightweight analysis where response speed matters more than depth.StepFun: Step 3.5 Flash (Free)
StepFun: Step 3.5 Flash (Free)
Model ID:
stepfun/step-3.5-flash:freeProvider: StepFunA fast free-tier model from StepFun. Good for quick verdicts where you want a different model perspective than the OpenRouter defaults.Google: Gemma 3 4B (Free)
Google: Gemma 3 4B (Free)
Model ID:
google/gemma-3-4b-it:freeProvider: Google AI StudioGoogle’s Gemma 3 4B instruction-tuned model, available for free via OpenRouter. A capable open-weight model for structured analysis tasks.Seeing the routed model
When you useopenrouter/auto or openrouter/free, OpenRouter selects the actual underlying model at request time. Once the analysis stream begins, the UI updates to display the real model name that was used — this comes from the model SSE event emitted during streaming.
This means the model shown before you submit may differ from the one shown in the results panel. The post-analysis model name is always the accurate one.
Passing a model via the API
If you are calling the API directly, use themodel query parameter:
Invalid model IDs
If you supply a model ID that is not in the allowed list, the platform falls back toopenrouter/free automatically — no error is returned and the analysis continues. The results will reflect whichever model was actually used.
All models are accessed via OpenRouter, so an OpenRouter API key is required regardless of which model you select. See Using your API keys for setup instructions.