01

Log in Start free

Worldwide model index · 24 models

Model Index

Search every major AI model in the world and its variants — Claude Fable / Opus / Sonnet / Haiku, the GPT family, Gemini tiers, Llama, Mistral, DeepSeek, Qwen and more. Click any model for capabilities, pros, cons, and pricing.

Compare them all in the Model Matrix

24 models · pricing indicative, reviewed July 2026

Claude Fable 5

Anthropic · USA · Text + vision

Anthropic's top flagship — sits above Opus for the hardest work.

$10 in / $50 outDetails

Claude Opus 4.8

Anthropic · USA · Text + vision

Heavy-duty reasoning and coding at the Opus tier.

$5 in / $25 outDetails

Claude Sonnet 5

Anthropic · USA · Text + vision

Near-Opus intelligence for coding and agents at Sonnet cost.

$3 in / $15 outDetails

Claude Haiku 4.5

Anthropic · USA · Text + vision

Fastest, cheapest Claude — built for simple tasks at high volume.

$1 in / $5 outDetails

GPT (OpenAI)Closed

GPT-5.6 Sol

OpenAI · USA · Text + vision + audio

OpenAI's flagship GPT-5.6 tier for the most demanding work.

$5 in / $30 outDetails

GPT (OpenAI)Closed

GPT-5.6 Terra

OpenAI · USA · Text + vision

The balanced everyday-work tier of GPT-5.6.

$2.5 in / $15 outDetails

GPT (OpenAI)Closed

GPT-5.6 Luna

OpenAI · USA · Text

Fastest, lowest-cost GPT-5.6 for high-volume workloads.

$1 in / $6 outDetails

GPT (OpenAI)Closed

GPT-5.5 Pro

OpenAI · USA · Text + vision

Maximum-reasoning tier — the priciest model on the ladder.

$30 in / $180 outDetails

Gemini 3.6 Flash

Google DeepMind · USA · Text + vision + audio + video

Google's fast flagship — strong coding, beats prior Pro on cost.

$1.5 in / $7.5 outDetails

Gemini 3.1 Pro

Google DeepMind · USA · Text + vision + audio + video

The 1M-context Pro tier for deep, grounded work.

$2 in / $12 outDetails

Gemini 3.1 Flash-Lite

Google DeepMind · USA · Text + vision

The cheapest active Gemini — high-volume, low-cost.

$0.25 in / $1.5 outDetails

Grok 4.5

xAI · USA · Text + vision

Unfiltered, on-trend, and grounded in real-time X.

$5 in / $25 outDetails

Llama 4 Behemoth

Meta · USA · Text + vision

Meta's largest open-weight model — frontier-adjacent quality.

Open / self-hostDetails

Llama 4 Maverick

Meta · USA · Text + vision

Balanced open-weight workhorse for production.

Open / self-hostDetails

Llama 4 Scout

Meta · USA · Text

Small, efficient open model with a long context window.

Open / self-hostDetails

Mistral Large 3

Mistral AI · France · Text

Mistral's flagship, now under Apache 2.0.

Open / self-hostDetails

Mistral Small 4

Mistral AI · France · Text

Efficient, cheap-to-run open model (Apache 2.0).

Open / self-hostDetails

Codestral 25.06

Mistral AI · France · Code

Coding-specialized open model.

Open / self-hostDetails

DeepSeek R1

DeepSeek · China · Text (reasoning)

Open-weight reasoning model with standout math scores.

Open / self-hostDetails

DeepSeek V3.2

DeepSeek · China · Text

Strong general open model with competition-grade results.

Open / self-hostDetails

Qwen 3.7 Max

Alibaba Cloud · China · Text + vision

Qwen's top model — leads on competition math.

Open / self-hostDetails

Qwen 3 235B

Alibaba Cloud · China · Text

Large open MoE model with strong benchmarks.

Open / self-hostDetails

Cohere Command R

Cohere · Canada · Text (RAG)

Enterprise retrieval-augmented generation and grounded search.

Open / self-hostDetails

PerplexityClosed

Perplexity (Sonar)

Perplexity AI · USA · Answer engine

Real-time, cited answers — research, not writing.

Open / self-hostDetails