Tencent · Released Jul 6, 2026Hy3
Tencent Hy reasoning model for coding, instruction following, and agent tasks.
- Capability
- —No ECI score yet
- Input
- $0.132per 1M tokens
- Output
- $0.53per 1M tokens
- Context
- 256K128K max output
What it costs
- Input
- $0.132per million tokens
- Output
- $0.53per million tokens
- Blended (3:1)
- $0.232cheaper than 77% of priced models
Typical monthly bills
| Monthly usage | Estimated cost |
|---|---|
| Side project2M input + 0.5M output tokens | $0.529 |
| Team assistant25M input + 5M output tokens | $5.95 |
| Production app250M input + 50M output tokens | $59.50 |
Median of 15 paid API providers on models.dev; Tencent also offers it free on Tencent TokenHub.
Independent benchmarks
Epoch AI has not published a Capabilities Index score for Hy3 yet.
The details
- Lab
- Tencent
- Released
- Jul 6, 2026
- Context window
- 256,000 tokens
- Max output
- 128,000 tokens
- Inputs
- Text
- Output
- Text
- Reasoning
- Adjustable effort high
- Tool calling
- Yes
- Structured output
- No
- Weights
- Open weights
- API model ID
hy3on Tencent TokenHub- Availability
- 18 API providerslisted on models.dev
What Tencent claims
Published by the lab at launch. Settings vary, so compare these only with care.
| Benchmark | Score | Setting | Source |
|---|---|---|---|
| SWE-Bench Verified | 78 resolved | — | Source |
| SWE-Bench Multilingual | 75.8 | highest reasoning effort | Source |
| SWE-Bench Pro | 57.9 | highest reasoning effort | Source |
| Terminal-Bench v2.1 | 71.7 | highest reasoning effort; 4h timeout; 500 episodes | Source |
| NL2Repo | 45.6 | highest reasoning effort; 250 turns; 12000s timeout | Source |
| DeepSWE | 28 | highest reasoning effort; 2h timeout | Source |
| BrowseComp | 84.2 | highest reasoning effort | Source |
| WideSearch | 76.4 | highest reasoning effort | Source |
| DeepSearchQA | 91 | highest reasoning effort | Source |
| MCP Atlas | 79.1 | highest reasoning effort; April 2026; 100 tool calls | Source |
| Toolathlon | 48.5 | highest reasoning effort | Source |
| APEX-Agents | 25.6 | highest reasoning effort | Source |
More from Tencent
About Hy3
How much does Hy3 cost?
Hy3 costs $0.132 input / $0.53 output per million tokens (median across 15 API providers; free on Tencent TokenHub). At a 3:1 input-to-output mix that is $0.232 per million tokens, cheaper than 77% of the 360 priced models we track.
What is the context window of Hy3?
Hy3 accepts up to 256,000 tokens per request and can write up to 128,000 tokens in one response.
How good is Hy3?
Epoch AI has not published a Capabilities Index score for Hy3 yet.
Is Hy3 open source?
Yes. Tencent publishes the weights, so it can be downloaded and self-hosted.
What inputs does Hy3 support?
Hy3 accepts text and replies in text. It is a reasoning model with high effort settings, supports tool calling.
When was Hy3 released?
Tencent released Hy3 on Jul 6, 2026.
What are the best alternatives to Hy3?
The closest current models from other labs on capability, price and release date are Qwen3.8 Flash (Alibaba (Qwen), $0.15 / $0.47), MiniCPM5-2B (OpenBMB, $0.124 / $0.743), Seed Character (ByteDance Seed, $0.119 / $0.297) and Viv Fast (Vivgrid, $0.13 / $0.40).