Fireworks

      The highest performance inference for open models.

      Start serverless in minutes, then scale to dedicated or reserved capacity without rewriting a line of code.

      • Fast, without cutting corners

        Up to 4x higher throughput with no hidden quality loss.

      • Serverless to reserved

        Pay per token, then scale to dedicated GPUs. No code rewrites.

      • Day-0 model support

        The newest open models, optimized the day they launch.

      Learn more in the docs →

      Authentication Section

      Create Account

      OR

      Already have an account?

      Log In

      By signing up, you agree to our Terms of service and Data Processing Agreement.