Is the web version a multimodal model?

If multi-modal models are not currently supported, is there any plan to add them in the future?
The code used in actual operation is the same as that of Gemma 3n E2B/E4B, yet errors still occur when using Gemma 4 E2B/E4B.

Image:

llm_inference_engine.cc:264] INVALID_ARGUMENT: Calculator::Open() for node "LlmVisionInferenceCalculator" failed: Image models could not be created.
=== Source Location Trace: ===
third_party/odml/infra/genai/inference/ml_drift/llm/llm_gpu_runner_manager.cc:509
third_party/odml/infra/genai/inference/calculators/llm_vision_inference_calculator.cc:276
third_party/mediapipe/framework/calculator_node.cc:603

Sound:

Uncaught (in promise) Error: Failed to create session: INVALID_ARGUMENT: Audio options should not be null when loading audio runner.
=== Source Location Trace: ===
third_party/odml/infra/genai/inference/ml_drift/llm/llm_gpu_runner_manager.cc:570
third_party/odml/infra/genai/inference/llm_engine.cc:980
third_party/odml/infra/genai/inference/llm_engine.cc:2681

Due to translation limitations, short sentences or brief texts often result in no response or incorrect translations.

Example:
Input Text :"Translate the following from en text into concise zh-Hant: "The code used in actual operation is the same as that of Gemma 3n E2B/E4B.".
Provide only the translated text. Do not include any additional explanations, commentary, or greetings."

Output:Output Text:
First time:Output Text: "" . //The response returned a null value.
Second time: Output Text:“The code used in operation is the same as that of Gemma 3n E2B/E4B.”
The correct result should be: “The code used in operation is the same as that of Gemma 3n E2B/E4B.” //Obtained duplicate input content

Every time it returns "Completed", repeating the translation several times will result in errors.
genai_bundle.js:1 Uncaught (in promise) Error: Cannot process because LLM inference engine is currently loading or processing.