Skip to content
TheChatArchiveThe Chat Archive
gpt-4o-search-preview-high· OpenAIHugging Face LMArena Search Arena 24K

find me a march 2025 overview of popular llm context length

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with gpt-4o-search-preview’s answer.
  1. User

    find me a march 2025 overview of popular llm context length

  2. Assistant

    As of March 2025, leading large language models (LLMs) have significantly expanded their context windows, enabling them to process and generate text with greater coherence over longer passages. Here's an overview of some prominent LLMs and their context lengths:

    ModelDeveloperRelease DateContext Window (Tokens)Notable Features
    Claude 3.7 SonnetAnthropicFeb 24, 2025200,000Hybrid reasoning model capable of "thinking" mode; knowledge cutoff extended to Oct 2024[1].
    Grok-3xAIFeb 17, 20251,000,000Integrated into X (formerly Twitter); processes real-time information; features "Think," "Big Brain," and "Deep Search" modes[1].
    Gemini 2.5 Pro ExperimentalGoogleMar 25, 20251,000,000Enhanced reasoning and coding capabilities; "thinking model" with chain-of-thought prompting; native multimodality[2].
    Llama 3.1 405BMetaEarly 2025128,000Open-source model with top-level abilities in general knowledge, multilingual translation, and tool usage; supports eight languages[3].
    DeepSeek-V3DeepSeekMar 24, 2025128,000Utilizes mixed-precision arithmetic for efficient sequence modeling; released under MIT license[4].

    These advancements reflect a trend toward increasing context lengths in LLMs, enhancing their ability to handle complex tasks and maintain context over extended interactions.

  3. User

    only use open models. Just a table with model name + 128k context columns

  4. Assistant

    As of March 2025, several open-source large language models (LLMs) support context windows of 128,000 tokens, enhancing their ability to process and generate text over extended passages. Below is a table highlighting these models:

    Model NameContext Length (Tokens)
    Llama 3.1 405B128,000
    Llama 3.1 70B128,000
    Llama 3.1 8B128,000
    Mistral NeMo 12B128,000
    Mistral Small 3.1 24B128,000
    Yarn-Mistral-7B-128K128,000
    DeepSeek-V3 671B128,000
    SEA-LION v3 8B128,000
    SEA-LION v3 70B128,000

    These models are openly available and offer extensive context lengths, facilitating improved performance in tasks requiring long-range coherence and understanding.

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