Current Instruments : Text Max, Text Prime & Text Standard

The Grid is a real-time spot market for AI inference. Today, three text-to-text Instruments are available for trading: Text Max, frontier models for deep reasoning, long context, and complex workflows. Text Prime, optimized for quality and long-form coherence at a high intelligence floor, and Text Standard, optimized for low latency and high throughput at a lower cost per Unit. Choosing between them depends on whether your workload prioritizes response quality or response speed - and since all markets use the same OpenAI-compatible Consumption API, you can switch between them by changing a single parameter.

Overview

The initial Text-to-Text Instruments available on The Grid are:

  • Text Standard: Optimized for speed and throughput.

    • Very low time to first token.

    • High streaming tokens per second.

    • Designed for short to medium outputs.

    • Intelligence floor that is good enough for many production workloads.

  • Text Prime: Optimized for quality and long form coherence.

    • Higher minimum intelligence benchmark score.

    • Larger maximum output size.

    • Accepts somewhat slower time to first token.

    • Designed for deep reasoning, long context, and complex tool use.

  • Text Max: Frontier models for deep reasoning, long context, and complex workflows.

    • Highest minimum intelligence benchmark score across all Text instruments.

    • 1M token context window and 128K token maximum output.

    • Accepts longer time to first token in exchange for frontier-class reasoning.

    • Designed for the most demanding workloads: extended analysis, multi-step synthesis, and complex tool use.

Which Instrument is right for you?

Choose Text Standard when:

  • You have tight latency budgets and need instant feeling UX.

  • You run many parallel calls where throughput and cost per token matter more than the last bit of reasoning quality.

  • Your prompts are short and you expect brief to moderate outputs.

  • You are doing routing, summarization, classification, or other transforms rather than complex planning.

Typical use cases:

  • Support chat assistants, help center bots, Slack and Discord helpers.

  • Email, meeting, and thread summarization, code diff explanations, document summaries.

  • RAG answers where context is already tight and responses should be concise.

  • Bulk text transforms such as classify, redact, extract fields, paraphrase, or translate short snippets.

  • Product surfaces that need instant feedback like autocomplete, inline suggestions, or hinting.

Last updated

Was this helpful?