Models & credits
Models
Which models each privacy mode offers, the presets in each mode, which TEE models currently verify, and how prices work.
On this page
Tacita gives you every top model in one place. The list depends on the mode you choose. Anonymous mode offers text, image and video models. TEE mode offers text models that run inside confidential-compute enclaves. Asking about images, image generation and video are Anonymous-only.
What each mode can do#
| Anonymous | TEE | |
|---|---|---|
| Text chat | 20 models | 12 open-weight models |
| Ask about an image (vision) | Yes, on vision models | No. End-to-end encryption currently carries text only. |
| Generate images | 13 models, 8 of which keep nothing | No |
| Generate video | 7 models, none of which keep your prompt or video | No |
| Tool calling | Yes, where the model supports it | No. Tool definitions would travel unencrypted. |
| Who can read the prompt | The model provider, which may log it | Only the attested enclave |
Counts are as of 28 October 2026 and change as models are added and retired. The app always shows the live list.
Presets#
Each mode has a short list of presets. All models shows everything else.
| Preset | TEE (text) | Anonymous (text) | Anonymous (image) | Anonymous (video) |
|---|---|---|---|---|
| Fast | GPT OSS 120B | DeepSeek V4.1 Flash | FLUX.2 Klein 4B | MiniMax H3 Max Turbo |
| Smart | GLM 5.3 | Claude Sonnet 5.5 | Seedream 4.5 | Flux 3 Max |
| Top | GLM 5.3 (see note) | Claude Opus 5.5 | Nano Banana (Gemini 2.5 Flash Image) | Grok Imagine 1.5 |
| Uncensored | Gemma 4 26B | DeepSeek V4.1 Flash Uncensored | — | — |
| Vision | — (TEE is text only) | GPT-6 Luna | — | — |
Notes:
- Top in TEE mode is the strongest model whose attestation currently verifies. Today that's GLM 5.3, the same as Smart. Kimi K3 becomes Top automatically if and when its attestation verifies. Until then you can still pick it from All models, without a Verified badge.
- TEE mode has no Vision preset. End-to-end encryption currently carries text only, so you can't attach images in TEE mode. Qwen3 VL 30B A3B is listed there, but only for text. To ask about a picture, use Vision in Anonymous mode, and remember that the model provider will see the image.
- Uncensored models refuse less often. The acceptable use rules still apply, and image and video prompts are still moderated.
All Anonymous models#
As of 28 October 2026:
| Type | Models |
|---|---|
| Text (20) | Claude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5, GPT-6 Luna, GPT-5.6 Luna, GPT-4o Mini, Gemini 3.1 Flash Lite, Grok 4.7, DeepSeek V4 Pro, DeepSeek V4.1 Flash, DeepSeek V4.1 Flash Uncensored, DeepSeek V3.2, Kimi K3, Kimi K2.7 Code, GLM 5.3, Qwen3.7 Flash, Qwen3 VL 235B A22B, MiniMax M3.1 Flash, Mistral Large 3, Llama 3.3 70B |
| Image (13) | Nano Banana (Gemini 2.5 Flash Image), GPT-5 Image Mini, Seedream 4.5, FLUX.2 Flex, FLUX.2 Klein 4B; not stored: Grok Imagine 2.0, Grok Imagine High Quality, Grok Imagine, Krea 2 Turbo, Hunyuan Image 3.0, Z-Image Turbo, Chroma, Anime (WAI) |
| Video (7) | Grok Imagine 1.5, Grok Imagine 1.5 Lite, Flux 3 Max, MiniMax H3 Max, MiniMax H3 Max Turbo, LongCat, LongCat Distilled |
Video models, and image models marked Not stored, run with providers that keep no copy of your prompt or the result. That's the provider's commitment, not hardware attestation as in TEE. Each video model can also start from an image you upload as the first frame.
TEE models#
All 12 TEE models are text models. They run in confidential-compute enclaves operated by Phala and NEAR AI. Attestation is checked for each model, and the result is live. The table shows the state on 27 October 2026.
| Model | Context | Enclave operator | Attestation |
|---|---|---|---|
| Kimi K3 | 1M | Phala | Does not currently verify |
| GLM 5.3 | 1M | Phala | Verifies |
| GLM 5.2 | 524K | Phala | Verifies |
| Kimi K2.6 | 262K | NEAR AI | Does not currently verify |
| Qwen 3.8 27B | 262K | NEAR AI | Does not currently verify |
| Qwen3 VL 30B A3B (text only in TEE) | 128K | Phala | Verifies |
| Gemma 4 26B A4B Uncensored | 64K | Phala | Verifies |
| Qwen 3.6 35B A3B FP8 | 32K | NEAR AI | Verifies |
| DeepSeek V4 Flash | 1M | NEAR AI | Verifies |
| GLM 5.3 Flash | 1M | NEAR AI | Verifies |
| GPT OSS 120B | 128K | Phala | Verifies |
| Qwen 2.5 7B | 32K | Phala | Verifies |
A model that doesn't verify can still be selected, but it shows no Verified badge. Without a passing attestation, you have no hardware proof that the model is running in a genuine enclave. Attestation explains what is checked.
Access#
Every model in both modes is open to anyone with credits, including Kimi K3, images and video. Your Main balance and the daily Bonus balance from a tier work the same way on any model. Your tier changes the size of your Bonus, not what you can use.
Beta models#
Some models carry a beta label, which means their availability is limited right now and requests fail more often. If a model can't be served, you get a "model unavailable" error and you aren't charged.
Prices#
You pay only for what you use. Every model has a transparent price per million tokens, per image or per video, shown in the picker. Pricing is the same for everyone, whatever your tier, and there are no subscriptions.
- Anonymous mode (text) runs on a secondary compute market. Prices are typically below each model's official list price, and every request is routed to the lowest available price for that model first. You're billed the actual price the request was served at, and never more than the official list price.
- TEE mode: you're charged exactly the model's official list price for actual usage.
- Images and video: you're charged the listed price.
See Credits.
The live list#
The model picker is the live catalogue. It shows each model's current price, a beta label where one applies, and, in TEE mode, whether its attestation currently verifies. Our server re-checks attestation for every TEE model about once an hour, and your browser checks it again before it shows a badge.