Fireworks said partner LangChain generates billions of agent-trace tokens a day, making frontier closed judges too costly. Fine-tuning a Qwen base on Fireworks matched GPT-5.5 / Opus-level judging at up to about 100x lower cost, and Fireworks is pushing its Training API for similar workflows.

Key Takeaways

  • ✓LangChain produces billions of agent-trace tokens daily; frontier closed judges are too expensive.
  • ✓Qwen fine-tuned on Fireworks matched GPT-5.5 / Opus-level judging at up to ~100x lower cost.
  • ✓Training API is positioned for builders to train similar low-cost judge/agent models.
ADSponsored