Skip to content
The Chat Archive
sonar-reasoning-pro-high LMArena Search Arena 24K

create a table comparing the input and output costs of these llms: sonnet 3.7 sonnet 3.5…

  1. User

    create a table comparing the input and output costs of these llms:

    sonnet 3.7 sonnet 3.5 gemini pro 2.5 gpt-4o gpt-4.1 Deepseek V3 Deepseek R1

  2. Assistant

    LLM Pricing Comparison: Input and Output Costs

    Below is a detailed comparison of input and output costs for the requested language models. The costs are standardized to price per million tokens for easy comparison.

    ModelInput Cost (per 1M tokens)Output Cost (per 1M tokens)Notes
    Claude 3.7 Sonnet$3.00$15.00Available through Amazon Bedrock and Google Cloud Vertex AI[5]
    Claude 3.5 SonnetNot specified in resultsNot specified in resultsPricing information not available in search results
    Gemini 2.5 Pro1.25(≤200Ktokens)<br/>1.25 (≤200K tokens)<br/>2.50 (>200K tokens)10.00(≤200Ktokens)<br/>10.00 (≤200K tokens)<br/>15.00 (>200K tokens)Long context queries (>200K tokens) have higher rates for all tokens[6]
    GPT-4o$5.00$15.00Based on 0.005per1Kinputtokensand0.005 per 1K input tokens and 0.015 per 1K output tokens[1]
    GPT-4.1Not specified in resultsNot specified in resultsPricing information not available in search results
    DeepSeek V30.14(cachemiss)<br/>0.14 (cache miss)<br/>0.07 (cache hit)$0.28Uses a Mixture-of-Experts architecture that enables cost efficiency[7]
    DeepSeek R1$0.14$0.55Open-source model under MIT license with competitive pricing compared to proprietary models[8]

    Key Observations

    • Most Affordable Input Processing: DeepSeek V3 with cache hit ($0.07 per 1M tokens) offers the lowest input cost among all models in the comparison.

    • Most Affordable Output Generation: DeepSeek V3 ($0.28 per 1M tokens) provides the cheapest output cost, making it a cost-effective option for text generation tasks.

    • Premium Pricing: Claude 3.7 Sonnet and GPT-4o have higher pricing tiers, reflecting their advanced capabilities and performance on benchmark tests.

    • Context Length Pricing: Gemini 2.5 Pro is notable for its tiered pricing structure based on context length, with a significant price increase for inputs exceeding 200K tokens.

    • Open Source Option: DeepSeek R1, despite being open-source under MIT license, still has API usage costs when accessed through the official API, though it can be self-hosted to eliminate token-based costs[7][8].

    This pricing information is current as of April 2025, but providers may update their pricing structures over time. For the most accurate and up-to-date pricing, it's advisable to check the official documentation of each provider.

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

Identical first prompt