Tencent Hy4 Flash: a 224B MoE with 1M context for $0.07 per million tokens
Tencent's slimmed-down Hy4 sibling: 224B total parameters with 12B active per token, native multimodal input and a 1M-token context, priced at $0.07 per million input tokens for everyday office work.
Why it matters
One-twelfth of Hy4's input price with the same 1M context, and in Tencent's blind test of 259 everyday work tasks it scored level with Claude Haiku 5.5 - the cheapest hosted frontier-tier model of the month.
What you could build with it
A small agency could build a multilingual document-drafting service for everyday office work on Hy4 Flash's $0.07-per-million pricing, offering bulk drafting and summarization to small businesses that could never afford frontier API bills.
Does it hold up?
Live on Tencent Cloud TokenHub and free in WorkBuddy through Nov 30; at $0.07 per 1M input tokens it is among the cheapest hosted models, but weights are only promised, not released, and availability outside China is unclear.
Built with Tencent Hy4 Flash
- Hy4 flash: 224B/12B MoE, 1M context, $0.07/1M inputthe9 · Spec sheet: 224B total (64B Engram), 12B active, 1M context, TokenHub pricing with a 50% launch discount through Nov 30.
- Tencent Launches Hy4 Flash, a 1M-Context Model at $0.07 per Million Tokenspasqualepillitteri · Launch numbers and the caveat that open weights are promised with no date and no license - for now it is API-only.
- Tencent launches lower-cost Hy4 flash AI modelcasrai · Sourced launch summary: Hy4 flash has its own hy4-flash identifier in Tencent Cloud's API guide, with OpenAI and Anthropic protocol support.
Learn more
First spotted on article: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Google AI Edge Gallery: offline speech recognition and synthesis on-deviceGoogle added fully offline speech recognition and text-to-speech to AI Edge Gallery, with ML Kit APIs on Android and…models · JEV 0.73Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.