Product change·Feb 23 → Mar 2, 2026·Week 10, 2026·Marketing & Content
Fireworks clarifies fine-tuning with reasoning traces increases token usage
What changed
Feature change
Added policy note explaining fine-tuning with reasoning traces increases tuned tokens due to conversation unrolling, with reference to SFT documentation
Before and after
Feb 23, 2026 vs Mar 2, 2026Pulse.
100%
What changed
Added policy note explaining fine-tuning with reasoning traces increases tuned tokens due to conversation unrolling, with reference to SFT documentation
Sign in to view the full comparison
See the before/after screenshots and every marked-up change.
Week 8, 2026 · Fireworks overhauls AI model lineup with GLM-5 and MiniMax M2 additionsFireworks Adds Kimi K2.5 Turbo at $0.99 per Million Input Tokens · Week 12, 2026
More from Fireworks
Full history →- Week 40, 2026·packagingFireworks adds GLM 5.3 Flash with 200K context, per-token training pricing
- Week 37, 2026·improvedFireworks rebuilds Serverless Training API catalog: drops Qwen 3.5 9B, adds DeepSeek V4 Flash and Muse Glimmer 30B
- Week 34, 2026·increaseFireworks: On-Demand GPU pricing up 11-30% from Sep 1; region surcharge to 1.5x
- Week 33, 2026·improvedFireworks: New GB300 GPU tier ($18/hr) + region premium pricing
- Week 32, 2026·packagingFireworks: New Serverless Training API added (Qwen 3.5 9B, 3.6 27B, Kimi K3)