Venice AI

Venice added Mercury 3, Inception's diffusion-based text model, and says it is available anonymously. Venice claims it runs 4x faster than Haiku 5.5 and 8x faster than GPT-6 Luna and beats both on agentic search and coding.

Venice announced that Black Forest Labs' FLUX 3 Max is live on its platform. Venice says it is tuned for speed, producing a 10-second video in about 12 seconds, and runs fully private.

Venice announced it has passed 5,000,000 users, saying the platform remains private, anonymous and unrestricted.

Venice announced that MiniMax M3.1 Flash is now available on its platform, including anonymously. Venice describes it as MiniMax's frontier model for coding, agentic work and visual reasoning, running at three times the speed of M3.

Venice says DeepSeek V4.1 Flash in the Venice app has been upgraded to DeepSeek's 1017 release, which Venice says outscores leading open-source and frontier models on most benchmarks. The upgraded model is also available through the Venice API as deepseek-v4-1-flash-1017.

Venice says six films are competing as finalists for its Lumara Grand Prize, worth $32,000 and 3.2M Venice credits. The winner will be revealed live on October 26 in New York; Venice is spotlighting the finalists, including Candy by @DiDi_OKK.

Venice added NVIDIA's open vision-language model Nemotron 3.5 Super VL, which can read images, documents, GUIs and video, with up to 10 images per request.

Venice added Hy4 Preview, Tencent's flagship reasoning model for coding, data analysis and research, supporting long-horizon reasoning, tool use and agentic workflows.

Venice made Meta's Muse Wonder, a frontier reasoning model for agentic work with sub-agent delegation and parallel tool calls, available on its platform. It also added Muse Spark, Meta's open-weight reasoning model aimed at coding, the same day.