# clm_8B6641F51FC4

subject: Fireworks AI (/record/fireworks-ai.md)  
predicate: feature_added  
value: Qwen, Llama, DeepSeek SFT support  
valid_from: 2025-06-13  
valid_to: null  
published: 2025-06-13  
recorded: 2026-10-29  
confidence: 0.70  
status: valid  
assertable: true  
superseded_by: null  
last_updated: 2026-10-31  
checked: 2026-10-31 · its source page states the value and the date  

> This now encompasses the Qwen 2/2.5/3 series, Phi 4, Gemma 3, the Llama family, as well as leading open-source MoE models like Deepseek R1 and V3, including, of course, any fine-tuned variants of those models!

What https://fireworks.ai/blog/supervised-fine-tuning-v2 says, word for word.

## Sources

- https://fireworks.ai/blog/supervised-fine-tuning-v2 (primary, Fireworks AI, published 2025-06-13)

---

AsOf is a temporal fact ledger: a structured, dated, sourced record of what is true and what changed, built for AI retrieval.