Tencent Hy4 Flash hits OpenRouter
Tencent has launched Hy4 Flash on OpenRouter, making its lightweight multimodal model globally accessible with a 1-million-token context window and image input capabilities. Running at roughly a quarter of flagship Hy4's parameter count, it delivers high-speed inference and tool calling at an introductory 50% discount through November 30 ($0.035/M input and $0.14/M output).
Tencent's rollout of Hy4 Flash on OpenRouter highlights an aggressive push to capture global developer mindshare with sub-dollar million-token multimodal inference. Offering a 1M context window alongside native vision at introductory prices below $0.04/M input raises the bar for budget developer models.
- –Quarter-scale efficiency: Shrinking parameter size to a quarter of the flagship model slashes latency and compute requirements while supporting up to 64K completion tokens.
- –Massive context retention: Maintaining the full 1,048,576-token context window allows cheap processing of extensive codebases, long document packages, and sustained agent conversations.
- –Production agent support: Native function calling, tool use, and structured JSON output support ensure seamless integration into existing developer agent frameworks.
- –Aggressive pricing pressure: At $0.035/M input and $0.14/M output through November 30, it undercuts most comparable long-context multimodal models on OpenRouter.
- –Frictionless global distribution: First-party deployment on OpenRouter bypasses domestic Tencent Cloud onboarding hurdles, giving international developers immediate API access.
DISCOVERED
2h ago
2026-11-16
PUBLISHED
3h ago
2026-11-16
RELEVANCE
AUTHOR
OpenRouter