Skip to content

Choosing chat models

Aethera lets you choose the model for each chat. The selected model controls answer style, reasoning depth, speed, and credit cost.

The model picker is in the chat composer. On a new chat it may briefly show Select model… while the catalog loads, then it falls back to your default model or Aethera’s default.

Aethera groups chat models into three access classes:

Standard

Class 1

Standard models are available to Free and paid users. Daily and bonus credits can be spent on Class 1 model work after monthly credits are used.

Advanced

Class 2

Advanced models are paid-plan models. They use monthly credits and are not eligible for daily or bonus credit spend.

Premium

Class 3

Premium models are the highest-cost models in the picker. Use them when quality matters more than speed or cost.

The live model catalog includes models from Anthropic, OpenAI, Google, xAI, Qwen, Kimi, Z.ai, DeepSeek, and MiniMax. The active picker currently includes:

ModelClassBest for
Claude Sonnet 4.62high-quality analysis, writing, planning
Claude Haiku 4.51fast Anthropic responses
Gemini 3.1 Flash Lite1efficient everyday work
Gemini 3 Flash1default fast general work
GLM-51general reasoning and drafting
MiniMax M2.51fast low-cost general work
Kimi K2.5 / K2.61research-style reasoning and long-form tasks
GPT-OSS 120b1open-model work with no provider cost basis
DeepSeek V4 Flash / Pro1efficient coding and analysis
Qwen 3.6 Flash / Plus1fast general reasoning
GPT-5.22advanced OpenAI reasoning
Gemini 3.1 Pro2stronger Google reasoning
Grok 4.3 / Grok 4.202xAI reasoning and conversational work
Qwen 3.6 Max2higher-quality Qwen work
Claude Opus 4.6 / 4.73premium Anthropic depth
GPT-5.4 / GPT-5.53premium OpenAI depth

Models marked legacy in old conversations may still appear in history, but the picker hides legacy models for new work.

Go to Chat from the sidebar.

Click the model selector in the composer. It appears near the persona selector and reasoning controls.

Use the picker to find a model by name or provider. If the picker is still loading, wait a moment and try again.

Choose the model you want. The next message uses that model. Existing assistant responses do not change.

To change the model used for new conversations:

  1. Open Settings → Account.
  2. Find Defaults.
  3. Choose a Default model.
  4. Save changes.

If the selected default model does not support reasoning, the Default thinking level field explains that reasoning is unavailable for that model.

Some models support configurable reasoning. When a model supports it, the composer shows a reasoning control beside the model/persona controls. If the current model does not support reasoning, Aethera disables or hides reasoning choices.

Use higher reasoning when you need careful multi-step analysis, planning, debugging, or synthesis. Use lower reasoning for quick drafts and simple questions.

Chat model costs are token-based:

  • input tokens come from your prompt, conversation context, files, citations, and tool results
  • output tokens come from the assistant’s answer and reasoning where applicable
  • each active model has its own input and output credit rate
  • tool calls and image actions can add separate credit costs

The model catalog is still loading. Wait a moment. If it remains empty, refresh the page.

Aethera may still be loading the selected model, you may be out of credits, or the message may not contain any usable input. Check Settings → Usage if you suspect a credit issue.

Models are periodically retired or moved to legacy status. Legacy models remain supported for old history and cost settlement, but they are hidden from the current picker.