Fireworks AI is listing Qwen3.8-Flash-Next FP8 as an experimental preview built on a Qwen4 architecture. The model is described as multimodal, with 125 billion parameters, 6 billion active parameters and an additional 51 billion n-gram embedding parameters.
Fireworks lists a 262,144-token context window and charges $1.20 per million tokens for both input and output. Its on-demand deployment option uses dedicated GPUs; Fireworks says it has no rate limits. The listing does not specify when the preview became available or who can access it.
Why it matters
For developers estimating inference costs, Fireworks lists equal input and output prices, but the page does not clarify preview access or availability timing.
