kapyn
Compare

AI model comparison

Compare the major AI models side by side — Claude, GPT, Gemini, Llama, and more — by context window, cost, modalities, and what each is genuinely best for. Calm, sourced, and free.

Cost is shown as a tier, not a live price — model pricing moves often, so always confirm current rates on the provider's page before you commit. Last updated 2026-07-01.

ModelTypeContextCostModalitiesOpenBest for
Claude OpusAnthropicFrontier200K$$$$text, visionThe hardest reasoning, long agentic tasks, and code that has to be right
Claude SonnetAnthropicBalanced200K$$text, visionThe default for most work — reads code like a senior engineer at a fair price
Claude HaikuAnthropicFast & cheap200K$text, visionHigh-volume, low-latency work where speed and cost matter most
GPT-4oOpenAIFrontier128K$$$text, vision, audioA strong all-rounder with native voice and vision in one model
GPT-4o miniOpenAIFast & cheap128K$text, visionCheap, fast, high-volume tasks like extraction and classification
OpenAI o-seriesOpenAIReasoning128K$$$$text, visionMulti-step reasoning, math, and problems that reward slow thinking
Gemini 2.5 ProGoogleFrontier1M$$$text, vision, audioHuge-context work — whole codebases, long documents, video reasoning
Gemini 2.5 FlashGoogleFast & cheap1MFree tiertext, vision, audioPrototyping and high-volume work — capable, fast, with a usable free tier
LlamaMetaOpen weights128KFree tiertext, visionRunning locally or self-hosting — private, free to run, no vendor lock-in
QwenAlibabaOpen weights128KFree tiertext, visionA strong open model for local use, especially good at coding for its size
DeepSeekDeepSeekReasoning128K$textOpen reasoning at a fraction of the cost of closed reasoning models
MistralMistral AIOpen weights128K$text, visionEfficient European open models for building without heavy lock-in
Claude OpusFrontier

The hardest reasoning, long agentic tasks, and code that has to be right

Context 200KCost $$$$text · vision

The default for most work — reads code like a senior engineer at a fair price

Context 200KCost $$text · vision
Claude HaikuFast & cheap

High-volume, low-latency work where speed and cost matter most

Context 200KCost $text · vision
GPT-4oFrontier

A strong all-rounder with native voice and vision in one model

Context 128KCost $$$text · vision · audio
GPT-4o miniFast & cheap

Cheap, fast, high-volume tasks like extraction and classification

Context 128KCost $text · vision

Multi-step reasoning, math, and problems that reward slow thinking

Context 128KCost $$$$text · vision

Huge-context work — whole codebases, long documents, video reasoning

Context 1MCost $$$text · vision · audio
Gemini 2.5 FlashFast & cheap

Prototyping and high-volume work — capable, fast, with a usable free tier

Context 1MFree tiertext · vision · audio
LlamaOpen weights

Running locally or self-hosting — private, free to run, no vendor lock-in

Context 128KFree tiertext · visionOpen weights
QwenOpen weights

A strong open model for local use, especially good at coding for its size

Context 128KFree tiertext · visionOpen weights
DeepSeekReasoning

Open reasoning at a fraction of the cost of closed reasoning models

Context 128KCost $textOpen weights
MistralOpen weights

Efficient European open models for building without heavy lock-in

Context 128KCost $text · visionOpen weights

Go deeper

The matrix is the quick answer. For the reasoning behind a pick, read Claude vs GPT-4o, Claude vs Gemini, or running open models locally. Every category above — Frontier, Balanced, Fast & cheap, Reasoning, Open weights — maps to a use case, not a leaderboard rank.

Embed this comparison

Free to use on your own site — paste this snippet where you want the live, auto-updating matrix to appear.

<iframe src="https://kapyn.app/embed/compare" width="100%" height="640" style="border:1px solid #222;border-radius:14px" title="AI Model Comparison by Kapyn" loading="lazy"></iframe>