Hy3
Hy3 is a 295B-parameter Mixture-of-Experts model from Tencent (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use.
Hy3Released 6 Jul 2026 (Jul 2026)Pricing
List prices at DeepInfra, in US dollars per million tokens.
| Tier | Input / 1M | Output / 1M | |
|---|---|---|---|
| Standard | $0.14 | $0.58 | |
| Cached input 75% off input | $0.035 | 75% off input |
What it costs a month
At DeepInfra list prices, 30 days.
- Chat assistant5,000 requests a day, 2K in / 0.5K out, cached$76.05
- RAG over documents2,000 requests a day, 8K in / 0.6K out, cached$62.88
- Coding agent1,000 requests a day, 40K in / 2K out, cached$115
- Overnight batch20,000 requests a day, 4K in / 0.4K out$475
Where to run it
6 hosts serve Hy3. The comparison is on blended price, 3 parts input to 1 part output.
Capabilities
Core
- Streaming
- Function calling
- JSON mode
- Structured outputs
Multimodal
- Vision
- Image input
- Audio
- Image generation
- Video
Advanced
- Reasoning
- Tool calling
- MCP
- Prompt caching
- Batch API
- Embeddings
- Takes in
- Text
- Produces
- Text
- Features
- Prompt caching
- Open weights
- License: Apache-2.0Runs in about 717.1 GB of GPU memory
Cheaper alternatives
Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.
Same model, cheaper host
No host sells it for less than DeepInfra.
Smaller Independent models
No cheaper Independent model keeps its key capabilities.
Other frontier labs
- 20% less
- 20% less
Open-weight picks
- 8% less
- 28% less
- 43% less
- 44% less
Model line
Newer and older versions, at each lab's own price.
- Nov 2026Hy4Latest$0.417 / $1.25
- Jul 2026Hy3This page$0.14 / $0.58
Call it
Using DeepInfra's API.
from litellm import completion
response = completion(
model="deepinfra/Hy3",
messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)Quick answers
How much does Hy3 cost?
On DeepInfra, Hy3 costs $0.14 per million input tokens and $0.58 per million output tokens. Cached input costs $0.035 per million tokens.
Which APIs offer Hy3?
Hy3 runs on 6 APIs: OpenCode Go, DeepInfra, GMI Cloud, Novita AI, AIHubMix and Kyma API.
What is Hy3's context window?
262K tokens, with up to 262K tokens in one reply.
Is Hy3 being retired?
No retirement date is scheduled for it on any API we track.
Is there a newer version of Hy3?
Yes. Hy4 is the newest model in this line, released on 4 Nov 2026.
What are cheaper alternatives to Hy3?
Models that cost less and keep most of what it can do: Claude Haiku 5.5 on Anthropic (20% less), GPT-6 Luna on OpenAI (20% less) and Qwen3.8 Omni Flash on Alibaba DashScope (8% less). Savings are on a mix of 3 input tokens to 1 output token.
When was Hy3 released?
Independent released it on 6 Jul 2026.
What is Hy3's API model id?
Use Hy3 with DeepInfra. Other hosts use their own ids: hy3 on OpenCode Go, hy3 on Novita AI, hy3 on AIHubMix, hy3 on Kyma API.
Sources
- DeepInfra pricing
- GMI Cloud pricing
- AIHubMix pricing
- Novita AI pricing
- OpenCode Go pricing
- LiteLLM model map
Raw catalog entry for Hy3 on DeepInfra
{
"max_output_tokens": 262144,
"max_input_tokens": 262144,
"input_cost_per_token": 1.4e-7,
"output_cost_per_token": 5.8e-7,
"cache_read_input_token_cost": 3.5e-8,
"supports_prompt_caching": true,
"litellm_provider": "deepinfra",
"mode": "chat",
"supports_function_calling": true,
"supports_tool_choice": true,
"supports_response_schema": true,
"supports_reasoning": true,
"supports_vision": false,
"source": "https://deepinfra.com/pricing"
}