Fireworks is letting anyone fine-tune Meta's 30B Muse Glimmer for free until Nov 30 — no credit card.
1️⃣ LoRA SFT, free via the UI or Serverless Training API — jobs must start before Dec 1, 00:00 PT
2️⃣ 30B dense model, Apache 2.0, built for agentic work — tool calling, 131k
Fireworks AIFireworks free Muse Glimmer fine-tuning
Free supervised fine-tuning (SFT) of Muse Glimmer 30B through November 30, no credit card required
Fireworks AI is an inference and model training platform for open models. Muse Glimmer 30B is free to train with supervised fine-tuning through November 30, kicking off its Free Weights Club.
LiveVerified Nov 3, 2026 against the official page
- Who qualifies
- Any Fireworks AI user, no credit card required
- How to claim
- Start an SFT job on Muse Glimmer 30B in the Fireworks app or API with Managed Training, or write your own loop with the Serverless Training API, before the end of November 30.
- Expiry
- 2026-11-30
Heads up. Only training is free - inference, including evaluation calls, is billed separately. Muse Glimmer 30B is deprecated from Fireworks public serverless inference, so serving a tuned model may need a dedicated deployment.
Timeline 2
Posts 1
Official sources
Added Nov 3, 2026 · Updated Nov 3, 2026, 08:22 UTC
