News Hub

ARCv3 demonstrates cross-region reinforcement learning for efficient model optimization.

Fireworks AI introduced Ember-1.

Fireworks AI introduced the Specialized Intelligence Index.

Gen-1 Slides generates Opus 5-level presentation decks at reduced cost.

DeepSeek V4 Pro achieves top SWE-Bench performance and reduces cost per task by 3x compared to Fable 5.

Fireworks AI post-trained Kimi K3 model with Harvey for long-horizon legal work applications.

Researchers reproduced J-Lens readouts on Kimi K3 and Qwen3.5-9B to detect silent signals before model output.

Fireworks AI offers performance stack combining Fireworks Models with Voyage AI embeddings.

Three tests to run before switching from LoRA to full fine-tuning.

Fine-tune embedding models from large language models.

Trilogy released Kimi K3, an open-weight cybersecurity model.

MiniMax M3 sparse attention optimization runs on NVIDIA Blackwell.

Gumloop achieved 7x scaling of open-weight model usage in three weeks using Fireworks AI.

Article reviewed seven open source large language models available in 2026.

An engineer completed one month of work in four days using GLM 5.2 Fast.

GLM 5.2 Fast model now available on Fireworks.

Factory increased open model usage 2-3x in six months using Fireworks infrastructure.

Fireworks AI offers open-source worker agents paired with closed-source advisor agents for cost-effective frontier AI.

GLM 5.2 launched on Fireworks inference platform on day zero.

Kimi K2.7 Code model available on Fireworks with improved agent performance and lower per-task costs.