Choose AI models with evidence.

Compare model capabilities, official prices, and public benchmarks. Key figures retain traceable sources, with regional and billing conditions kept explicit.

DATA SNAPSHOT · 2026-10-21

Choose AI models with evidence.

Compare model capabilities, current prices, and public evaluations. Follow every key figure back to its source, with regions and billing terms shown separately.

Browse models↗Compare up to three→

DATA SNAPSHOT

Artificial Analysis Intelligence Index

↗

Index points · 0–60 scale; not a percentage.

OpenAI · GPT-6 Astra53

max

Anthropic · Claude Opus 5.558

max with fallback

Google · Gemini 4 Argon53

high

DeepSeek · DeepSeek-V4.1 Flash45

max

Alibaba Cloud Qwen · Qwen3.8 Max45

Not specified by leaderboard

xAI · Grok 4.746

xhigh

Checked 2026-10-21Artificial Analysis leaderboard ↗

Providers coveredOpenAIClaudeGeminiDeepSeekQwenGrok

Official sources recorded

Selected models

Selected API models with source-linked prices and reasoning settings.

Browse models ↗

OpenAIFlagship

GPT-6 Astra

For complex reasoning and coding tasks.

API IDgpt-6-astra

Context 1,050,000 tokens · Max output 128,000 tokens

Standard price · Official API list price · per 1M tokens

Input$10

Cached inputNo comparable data

Output$50

Artificial Analysis Intelligence Index: 53 · max

Official model page ↗Checked 2026-10-21

AnthropicFlagship

Claude Opus 5.5

Well suited to long-running agentic coding and knowledge work.

API IDclaude-opus-5-5

Context 1,000,000 tokens · Max output 128,000 tokens

Standard price · Claude API · per 1M tokens

Input$4

Cached inputNo comparable data

Output$20

Artificial Analysis Intelligence Index: 58 · max with fallback

Official model page ↗Checked 2026-10-21

GoogleFlagship

Gemini 4 Argon

For complex software engineering, math, science questions, and multi-step agent tasks.

API IDgemini-4-argon

Context 1,048,576 tokens · Max output 65,536 tokens

Introductory price (through 2026-11-30) · Gemini API Standard tier · per 1M tokens

Input$2

Cached input$0.1

Output$10

Artificial Analysis Intelligence Index: 53 · high

Official model page ↗Checked 2026-10-21

DeepSeekHigh throughput

DeepSeek-V4.1 Flash

Supports thinking and non-thinking modes; API pricing varies by peak or off-peak period and cache hit status.

API IDdeepseek-flash

Context 1,000,000 tokens · Max output 384,000 tokens

Peak hours · Official API · per 1M tokens

Input$0.3

Cached input$0.006

Output$1.2

Artificial Analysis Intelligence Index: 45 · max

Official model page ↗Checked 2026-10-21

Alibaba Cloud QwenFlagship

Qwen3.8 Max

Supports thinking and non-thinking modes; available regions and prices depend on the Model Studio deployment region.

API IDqwen3.8-max

Context 1,000,000 tokens · Max output No comparable data

Real-time API calls · Global deployment (official US/Virginia table) · per 1M tokens

Input¥12

Cached inputNo comparable data

Output¥36

Artificial Analysis Intelligence Index: 45 · Not specified by leaderboard

Official model page ↗Checked 2026-10-21

xAIFlagship

Grok 4.7

The text API offers a 500K context window; requests at the 200K input threshold are billed at long-context rates.

API IDgrok-4.7

Context 500,000 tokens · Max output No comparable data

Short context (under 200K) · Global endpoint · per 1M tokens

Input$2

Cached input$0.5

Output$6

Artificial Analysis Intelligence Index: 46 · xhigh

Official model page ↗Checked 2026-10-21

Anran AI Guide is independent of model providers. Prices and availability can change.