Hy3 preview
Hy3 preview is a high-efficiency Mixture-of-Experts model from Tencent designed for agentic workflows and production use. It supports configurable reasoning levels across disabled, low, and high modes, allowing it to balance speed and depth depending on the task, while delivering strong code generation and reliable performance across multi-step, real-world workflows.
- Organization
- tencent
- Family
- Hy
- Providers
- 5
- Context
- 256,000
- Output limit
- 64,000
- Knowledge
- —
- Release
- 2026-04-20
- Updated
- 2026-04-20
- Weights
- Open
- Input
- text
- Output
- text
- Capabilities
- Tools, Reasoning, Temperature
Providers
Compare every cataloged offer for this model, including its provider, route, token limits, price, and capabilities.
| Provider | Model ID | Context | Output | Input / output · 1M | Reasoning | Tools | Structured | Details |
|---|---|---|---|---|---|---|---|---|
| AIHubMix | hy3-preview | 256,000 | 128,000 | $0.17 / $0.566661 | Yes | Yes | Yes | View |
| Kilo Gateway | tencent/hy3-preview | 262,144 | 235,929 | $0.18 / $0.60 | Yes | Yes | No | View |
| GMICloudvia OpenRouter | tencent/hy3-preview bf16 | 262,144 | 235,929 | $0.18 / $0.60 | Yes | Yes | No | View |
| Tencent TokenHub | hy3-preview | 256,000 | 64,000 | $0.00 / $0.00 | Yes | Yes | — | View |
| ZenMux | tencent/hy3-preview | 256,000 | 64,000 | $0.172 / $0.572 | Yes | Yes | — | View |
Prices and limits apply to each offer. A dash means the value is not provided. OpenRouter offers show the hosting provider and routing channel separately.
Pricing
Compare token rates and cache charges across available routes and hosting configurations.
Lowest listed input and output prices across the offers above; they may belong to different providers. These are base rates, before conditional pricing, discounts or additional fees.
tencent/hy3-previewObserved - Input
- $0.18
- Output
- $0.60
- Cache read
- $0.06
Hosting prices through OpenRouter All endpoints and cache rates
| Endpoint | Input / 1M | Output / 1M | Cache read / 1M | Cache write / 1M |
|---|---|---|---|---|
| GMICloudgmicloud/bf16 | $0.18 | $0.60 | $0.06 | — |
Each row is a hosting configuration. Open an offer in Providers for its full pricing conditions and additional fees. Endpoint observations may differ from the routing catalog's base price.
Performance
Review reported latency and output-throughput percentiles from the latest observed rolling window.
Time to first token · milliseconds
All latency percentiles 1 endpoints
| Endpoint | P50 | P75 | P90 | P99 |
|---|---|---|---|---|
| GMICloudgmicloud/bf16 | 2,421 ms | 2,929.5 ms | 5,872.8 ms | 15,821.76 ms |
Output throughput · tokens per second
All throughput percentiles 1 endpoints
| Endpoint | P50 | P75 | P90 | P99 |
|---|---|---|---|---|
| GMICloudgmicloud/bf16 | 93 t/s | 102 t/s | 111 t/s | 157.96 t/s |
Percentiles describe the observed request distribution. These are snapshots of a rolling window, not a historical time series.
Uptime
Compare successful request rates across the latest 5-minute, 30-minute, and 24-hour observations.
Reported by OpenRouter at the time of collection; rate-limited requests are excluded. A dash means no measurement was reported. This is not a live availability check.
Benchmarks
Review attributed evaluation results on each source's original scale.
tencent/hy3-previewObserved Sources: Artificial Analysis and Design Arena, via OpenRouter. Scores retain their original scales and are not OpenAI Suite ratings.
Detailed evaluations
| Evaluation | Configuration | Source | Results | Details |
|---|---|---|---|---|
| Composite indices | Hy3tencent/hy3-preview-20260421 | Artificial Analysis | Intelligence 25.3 · Coding 58.8 · Agentic 24.1 | Details |
Sources: Artificial Analysis, Design Arena and OpenRouter. Evaluation configurations and scales differ; results are not interchangeable. Updated Oct 4, 2026.
Apps & session costs
Compare observed median session costs for applications using this model, grouped by conversation length.
| Application | Model route | Turns | Median cost / session | Details |
|---|---|---|---|---|
| Hermes Agent | tencent/hy3-preview-20260421 | 2–9 turns | $0.012224 | Details |
| Hermes Agent | tencent/hy3-preview-20260421 | 10–49 turns | $0.095327 | Details |
Source: OpenRouter, as of Oct 2, 2026. Published under CC BY 4.0. These are observed costs per session, not token prices or a forecast for your workload. Apps are compared separately.
Activity
Daily prompt and completion tokens reported for this model's routes in OpenRouter's top 50. Missing days have no published value.
112,449,515,842 tokens
37,858,484,337 tokens
Daily observations · 2 days
| Date (UTC) | Reported tokens | Reported routes |
|---|---|---|
| Jul 8, 2026 | 37,858,484,337 | 1 |
| Jul 7, 2026 | 112,449,515,842 | 1 |
Source: OpenRouter (openrouter.ai/rankings), as of Oct 4, 2026. CC BY 4.0. Totals include only individually published routes. Tokenizers differ by provider; the aggregated “other” category is never assigned to a model.
Technical details & API
Explore specifications, supported parameters, reasoning controls, and API identifiers for each OpenRouter route.
tencent/hy3-previewObserved - Context
- 262,144
- Maximum output
- 235,929
- Tokenizer
- Other
- Input
- text
- Output
- text
- Moderation
- No
- Added to OpenRouter
- 2026-04-22
https://openrouter.ai/api/v1tencent/hy3-previewtencent/hy3-preview-20260421tencent/Hy3-previewReasoning
- Required
- No
- Enabled by default
- Yes
- Default effort
- high
- Supported efforts
- high, low, none
Supported parameters
include_reasoningmax_tokensreasoningreasoning_effortseedtemperaturetool_choicetoolstop_p
Default parameters
- top p
- 1
- temperature
- 0.9
Frequently asked questions
Answers based on the model metadata and serving offers currently in the catalog.
How much does Hy3 preview cost?
Listed rates start at $0.00 per million input tokens and $0.00 per million output tokens. Prices vary by provider and configuration; see Pricing for details.
Which providers offer Hy3 preview?
There are 5 cataloged offers across 5 providers. Compare model IDs, prices and limits in Providers.
What is the context window?
The cataloged model context is 256,000 tokens. Each serving endpoint may apply a different limit.
Does it support tools and structured output?
Tool calling: Yes. Structured output: Not reported. Support can vary by endpoint.
Resources
Catalog history · 2 recorded revisions
Metadata observations since this model was first cataloged. These are not model release versions.
- ·
cd89873223ce - ·
fa75628a21d5