Companion Model Compatibility

This table tracks which Companion features work with each provider and model. Only models with reported test data are listed.

Some results were generated from automated tests, so individual experiences may vary. If you believe something is inaccurate, please submit a pull request to update the table.

Linked notes are not tracked as a separate compatibility column because they are sent as plain text context.

Feature key

• 🟢 Fully functional — 🟡 Partially functional — 🔴 Non-functional
• Chat — Conversational responses about your note
• Edit — Proposes changes you can accept into your document
• Flashcards — AI flashcard drafting
• Tests — AI test generation
• Attachments — File attachments (DOCX, PPTX, PDF)
• Image — Image attachment support

DeepSeek

Model	Chat	Edit	Flashcards	Tests	Attachments	Image	Comments
deepseek-chat	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
deepseek-reasoner	🟢	🟢	🟢	🟢	🟢	🔴	No image support.

OpenAI

Model	Chat	Edit	Flashcards	Tests	Attachments	Image	Comments
gpt-5	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.
gpt-5-mini	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.
gpt-5-nano	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.
gpt-4.1	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.
gpt-4.1-mini	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.

OpenRouter

The following models were all tested via OpenRouter. They are grouped by provider for readability.

Amazon Nova

Model	Tier	Chat	Edit	Flashcards	Tests	Attachments	Image	Comments
amazon/nova-lite-v1	Paid	🟢	🟢	🟢	🔴	🟢	🟢	Tests unreliable.
amazon/nova-micro-v1	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
amazon/nova-premier-v1	Paid	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.
amazon/nova-pro-v1	Paid	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.

Meta Llama

Model	Tier	Chat	Edit	Flashcards	Tests	Attachments	Image	Comments
meta-llama/llama-3.1-70b-instruct	Paid	🟢	🟢	🟢	🟢	🔴	🔴	No attachment or image support.
meta-llama/llama-3.2-11b-vision-instruct	Paid	🟢	🟢	🟢	🟢	🟡	🔴	Only PPTX works; no image support.
meta-llama/llama-3.3-70b-instruct	Paid	🟢	🟢	🟢	🟢	🟡	🔴	PDF not supported; no image support.
meta-llama/llama-4-maverick	Paid	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.
meta-llama/llama-4-scout	Paid	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.

Mistral

Model	Tier	Chat	Edit	Flashcards	Tests	Attachments	Image	Comments
mistralai/mistral-large	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
mistralai/mistral-medium-3.1	Paid	🟢	🔴	🟢	🟢	🟢	🟢	Edit not supported.
mistralai/mistral-nemo	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
mistralai/mistral-small-3.2-24b-instruct	Paid	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.
mistralai/pixtral-large-2411	Paid	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.

Nvidia

Model	Tier	Chat	Edit	Flashcards	Tests	Attachments	Image	Comments
nvidia/llama-3.3-nemotron-super-49b-v1.5	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
nvidia/nemotron-3-super-120b-a12b	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.

Qwen

Model	Tier	Chat	Edit	Flashcards	Tests	Attachments	Image	Comments
qwen/qwen3-235b-a22b	Paid	🟢	🟢	🟢	🔴	🟢	🔴	Tests unreliable; no image support.
qwen/qwen3-32b	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
qwen/qwen3-max	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
qwen/qwen-plus	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
qwen/qwq-32b	Paid	🟢	🔴	🟢	🟢	🟢	🔴	Edit not supported; no image support.

Other Providers

Model	Tier	Chat	Edit	Flashcards	Tests	Attachments	Image	Comments
baidu/ernie-4.5-300b-a47b	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
bytedance-seed/seed-1.6	Paid	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.
cohere/command-a	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
inception/mercury-2	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
inflection/inflection-3-productivity	Paid	🟢	🟢	🟢	🟢	🟡	🔴	PPTX not supported; no image support.
liquid/lfm-2-24b-a2b	Paid	🟢	🟢	🔴	🟢	🟢	🔴	Flashcards unreliable; no image support.
microsoft/phi-4	Paid	🟢	🟢	🟢	🟢	🟡	🔴	PPTX/PDF can be unreliable; no image support.
minimax/minimax-m2.7	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
moonshotai/kimi-k2	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
moonshotai/kimi-k2.5	Paid	🟢	🟢	🟢	🟢	🟢	🟢	Fully functional.
nousresearch/hermes-4-405b	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
nousresearch/hermes-4-70b	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
stepfun/step-3.5-flash	Paid	🟢	🔴	🔴	🟢	🟢	🔴	Edit and flashcards not supported; no image support.
tencent/hunyuan-a13b-instruct	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
writer/palmyra-x5	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
xiaomi/mimo-v2-pro	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
z-ai/glm-4.6	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
z-ai/glm-4.7	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
z-ai/glm-5	Paid	🟢	🟢	🟢	🟢	🟢	🔴	No image support.
arcee-ai/trinity-large-preview:free	Free	🟢	🟢	🟢	🟢	🟢	🔴	No image support.

Routers

These are meta-models that route your request to another model at runtime. Results vary depending on which underlying model is selected.

Model	Tier	Chat	Edit	Flashcards	Tests	Attachments	Image	Comments
openrouter/auto	Paid	🟢	🟢	🟢	🟢	🟢	🟢	Routes to best available model.
openrouter/free	Free	🟡	🟡	🟡	🟡	🟡	🟡	Depends on which free model is chosen.

Models Not Supported

Model	Notes
arcee-ai/maestro-reasoning	All features failed.
arcee-ai/virtuoso-large	All features failed.
minimax/minimax-m1	All corefeatures failed.
rekaai/reka-flash-3	All core features failed.

How to test and contribute

If you test a provider/model that is not listed, you can submit a pull request to add it.

What to test

1. Chat — Open Companion and ask it to review your note. This should provide a chat response.
2. Edit — Open Companion, after asking it to review your note, ask it to implement changes. If it generates changes that you can accept into your document then that is correct.
3. Generate Flashcards — Open Companion and ask it to generate flashcards from a note. Confirm cards appear and can be inserted into your note.
4. Generate Tests — Open the Tests feature and generate a short test from a note. Confirm questions appear and are answerable. Complete the test, and confirm that the test is appropriately graded.
5. File Attachments — Embed or link a PDF or image in a note, enable Include embedded attachments in settings. Confirm the model processes the attachment without errors.

How to submit your results

1. Fork the repository and edit site/src/content/docs/Companion-Model-Compatibility.md.
2. Add a new row for the provider and model you tested, then mark each feature with 🟢 (fully functional), 🟡 (partially functional), or 🔴 (non-functional).
3. Add a short comment if there are caveats.
4. Open a pull request with the provider, model, and a brief summary of what you tested.

Last modified: 19/04/2026