$0.13 / $0.53per 1M tokens in / out
$0.033cache read · −75%
$0.072per task at thinking?
#13on OpenRouter by tokens · 3.1T / week
Price sheetper 1M tokens
| Input | $0.13 | list |
| Output | $0.53 | |
| Cache read | $0.033 | −75% vs input |
| Context · max output | 256K · 128K | tokens |
Where to buy7 providers · per 1M tokens
| Provider | Input | Output | Context |
|---|---|---|---|
| DeepInfra | $0.13 | $0.53 | — |
| Tencentlab | $0.13 | $0.53 | 262K |
| GMICloud | $0.14 | $0.58 | 262K |
| Novita | $0.14 | $0.58 | — |
| Phala | $0.15 | $0.64 | 262K |
| AtlasCloud | $0.20 | $0.80 | 262K |
| Featherless | $1 | $4 | — |
In Tencent's line-upper 1M in / out
| Hy MT2 1.8B | $0.044 / $0.18 | −67% |
| Hy MT2 30B A3B | $0.074 / $0.29 | −44% |
| Hy MT2 7B | $0.074 / $0.29 | −44% |
| Hy3 | $0.13 / $0.53 | this model |
| Hunyuan A13B Instruct | $0.14 / $0.57 | +7% |
| Hy4 preview | $0.83 / $2.50 | 5.4× |
Cheaper for the same workvs Hy3
| GPT-6 Luna max | -6% | score +51% | $0.068 |
| MiMo v2.6 Flash thinkingOpen | -13% | score +50% | $0.062 |
| Agnes 2.5 Pro Beta | -35% | score +39% | $0.10 / $0.30 |
| Claude Haiku 5.5 medium | -34% | score +36% | $0.047 |
| GPT-5.6 Luna high | -39% | score +27% | $0.044 |
Pricing questions
How much does Hy3 cost?›
$0.13 per 1M input tokens and $0.53 per 1M output tokens. Cached input is $0.033. One Intelligence Index task costs about $0.072 at the default effort.
Is Hy3 free to use?›
There is no free API tier, but the weights are open: you can run it on your own hardware for the cost of compute.
How much does prompt caching save?›
Cache reads cost $0.033 per 1M, 75% below the input price.
Is Hy3 cheaper than Hy3 preview?›
Yes: $0.13 / $0.53 against $0.18 / $0.60, 19% less per token.
What is a cheaper alternative to Hy3?›
GPT-6 Luna at max effort costs -6% per task with a score +51% on Humanity's Last Exam.