Model comparison

Kwoat gives you access to 10+ AI models. Here's how to choose — or just use Auto mode.

Auto mode (recommended)

Auto mode routes your prompt to the best model automatically based on task type, complexity, and your plan. For most users this is the best choice.

Model tiers

Standard tier — fastest, most affordable

Best for: everyday chat, simple questions, drafting, quick tasks.

ModelProviderContextVisionBest for
Gemini 2.5 FlashGoogle1M tokensFast general tasks, web search, large context
GPT-4o MiniOpenAI128K tokensGeneral chat, quick answers
Mistral Small 4Mistral AI262K tokensFast, vision-capable, European data
Llama 3.3 70BMeta66K tokensOpen source, coding, reasoning — no vision

Advanced tier — more capable, moderate cost

Best for: complex analysis, long documents, nuanced writing, detailed code.

ModelProviderContextVisionBest for
GPT-5.4OpenAI1M tokensComplex reasoning, vision, coding
Gemini 3.1 ProGoogle1M tokens✓ + audioVery long documents, audio analysis, 1M context
Grok 4.3xAI2M tokensLargest context window — 2M tokens for long research

Premium tier — most powerful

Best for: the hardest problems, research, complex multi-step reasoning.

ModelProviderContextVisionBest for
Claude Sonnet 4.6Anthropic1M tokensWriting, analysis, coding, nuanced tasks
Claude Opus 4.7Anthropic200K tokensHardest reasoning problems, deep research · 10/day Pro, 50/day Pro Max

Specialised models

ModelTypeCostNotes
Nano Banana 2Image generation$0.15/imageHigher quality, contextual understanding
FLUX.2 KleinImage generation$0.04/imageFast, cost-efficient
Kling 2.6 ProVideo generation$1.50 (5s) · $2.50 (10s)Cinematic video with audio

Which model should I use?

For writing

Claude Sonnet 4.6 consistently produces the most natural, nuanced writing. For quick drafts, Gemini 2.5 Flash is fast and excellent value.

For coding

GPT-5.4 and Claude Sonnet 4.6 are both excellent.Llama 3.3 70B is a strong option for standard coding tasks (note: no vision support). For the hardest architecture or debugging problems, Claude Opus 4.7 is the strongest choice.

For research and analysis

Claude Opus 4.7 or GPT-5.4 for deep research. Enable web search to get current information from any model. For very long source documents, Grok 4.3 (2M context) handles the largest inputs.

For long documents

Grok 4.3 has the largest context window at 2M tokens.Gemini 2.5 Flash and Gemini 3.1 Pro also support 1M tokens each.

For image analysis

All models except Llama 3.3 70B support vision. Gemini 3.1 Pro additionally supports audio input. For generating images, use Nano Banana 2 (quality) or FLUX.2 Klein (speed).

Plan access

PlanStandardAdvancedPremiumImageVideoMissions
Free
Pay As You Go✓ wallet✓ wallet
Pro✓ 20/mo✓ wallet
Pro Max✓ 100/mo✓ 5/mo incl.✓ 10/mo