Skip to content
All Models

INTELLECT 3

prime-intellect/intellect-3

INTELLECT 3 is a 100B+ Mixture-of-Experts language model from Prime Intellect. Through AI Gateway it runs with a context window of 131.1K tokens at $0.20 input and $1.10 output per million tokens. Model type: chat.

ReasoningTool Use
index.ts
import { streamText } from 'ai'


const result = streamText({
  model: 'prime-intellect/intellect-3',
  prompt: 'Why is the sky blue?'
})

What To Consider When Choosing a Provider

  • Zero Data Retention

    AI Gateway supports Zero Data Retention for this model via direct gateway requests (BYOK is not included). To configure this, check the documentation.

    Authentication

    AI Gateway authenticates requests using an API key or OIDC token. You do not need to manage provider credentials directly.

Compare list prices of $0.20 input and $1.10 output against your expected volume. Cached input: N/A.

Yes, Zero Data Retention is available for this model

Yes, a no-prompt-training provider is available for this model

When to Use INTELLECT 3

Best For

  • Math and scientific reasoning:

    Published benchmark results for the 100B+ MoE class matter

  • Software engineering tasks:

    Bug fixes, feature implementation, and codebase navigation in agentic workflows

  • Research and analysis:

    Needs multi-step reasoning across technical domains

Consider Alternatives When

  • Fastest inference required:

    MoE models at this scale need more compute than smaller dense models

  • Simple classification tasks:

    A lighter model costs less for straightforward extraction

  • Vision or multimodal input:

    INTELLECT 3 doesn't offer image, audio, or video input

  • Long-term support guarantees:

    Production stability depends on a vendor you already trust

Conclusion

INTELLECT 3 is a 100B+ MoE model for math, code, and reasoning workloads. AI Gateway exposes INTELLECT 3 with routing, retries, and failover across parasail. This model is available through AI Gateway

FAQ

It's a 100B+ Mixture-of-Experts model built on the GLM-4.5-Air base. MoE routes each input to a subset of expert parameters, so inference cost stays lower than a dense model with the same total parameter count.

Set your API key in your AI Gateway project settings. Use prime-intellect/intellect-3 in API calls. AI Gateway handles provider routing, retries, and failover across parasail. Try requests in the playground: https://ai-sdk.dev/playground/parasail:parasail-primeintellect3.

Prime Intellect published benchmark results for this size class on math, code, science, and reasoning. MoE activates a subset of experts per forward pass, which changes the cost profile versus dense models.