Skip to content
TheChatArchiveThe Chat Archive
DeepSeekdeepseek-v3-0324· DeepSeekHugging Face LMArena Human Preference 140K

what are the most expensive models available right now? especially from google and…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with gpt-4.1-2025-04-14’s answer.
  1. User

    what are the most expensive models available right now? especially from google and anthropic

  2. DeepSeek Assistant

    As of mid-2024, the most expensive AI models from Google and Anthropic are their highest-tier, state-of-the-art (SOTA) models, which come with premium pricing due to their advanced capabilities, large parameter sizes, and high computational costs. Here’s a breakdown:


    Most Expensive Google Models

    1. Gemini 1.5 Ultra (or Pro 1.5 in some versions)

      • Expected Cost: Likely the most expensive Google model (exact pricing not fully public yet).
      • Context Window: Up to 1 million tokens (for select customers).
      • Use Case: High-end enterprise applications, complex reasoning, and long-context tasks.
      • Pricing:
        • Google has not fully disclosed pricing for Gemini 1.5 Ultra, but it is expected to be significantly more expensive than Gemini 1.0 Pro (which costs ~$0.0025 per 1K characters for input).
        • The 1M token context feature is likely very costly (possibly $10+ per long query in early access).
    2. PaLM 2 (Legacy, but still expensive for large-scale use)

      • Cost: ~0.002per1Kcharacters(input)and 0.002 per 1K characters (input) and ~0.004 (output) via Vertex AI.
      • Use Case: Large-scale enterprise AI applications.

    Most Expensive Anthropic Models

    1. Claude 3 Opus

      • Cost: 15permillioninputtokens,15 per million input tokens, 75 per million output tokens (as of March 2024).
      • Context Window: 200K tokens (standard).
      • Use Case: The most advanced model from Anthropic, designed for highly complex reasoning, research, and enterprise applications.
      • Why Expensive?
        • Outperforms GPT-4 in some benchmarks.
        • High computational demand due to its size and capabilities.
    2. Claude 3 Sonnet (Mid-tier, but still costly for heavy usage)

      • Cost: 3permillioninputtokens,3 per million input tokens, 15 per million output tokens.
      • Use Case: A balance between cost and performance, but still expensive for large-scale deployments.

    Comparison of Costs (Per Million Tokens)

    Model (Most Expensive)Input CostOutput CostNotes
    Claude 3 Opus$15$75Most expensive Anthropic model
    Gemini 1.5 UltraLikely $10+Likely $30+Not fully public yet
    GPT-4-turbo (OpenAI)$10$30For reference
    Claude 3 Sonnet$3$15Mid-tier but still costly

    Why Are These Models So Expensive?

    • Compute Intensity: Larger models require massive GPU/TPU clusters to run.
    • Long Context: Models like Gemini 1.5 (1M tokens) and Claude 3 (200K tokens) are optimized for long documents, which increases costs.
    • Enterprise Demand: High-performance models are priced for businesses, not casual users.

    Which One Should You Use?

    • If you need the absolute best performance and cost is no concern → Claude 3 Opus or Gemini 1.5 Ultra.
    • If you need a balance of cost and quality → Claude 3 Sonnet or Gemini 1.5 Pro.

    Would you like recommendations based on a specific use case (e.g., coding, long-document analysis, etc.)?

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.