Hy3: Specifications & Sources

Specifications and access

Specification	Value and source
Weights	Open weightsSource
Access	API, LocalSource
Parameters	295BSource
Active parameters	21BSource
Context window	262,144 tokensSource
License	Apache-2.0Source
Parameters	295 BSource
Active parameters	21 BSource
MTP parameters	3.8 BSource
Architecture	sparse MoE, 80 transformer layers plus one MTP layer, 192 experts with top-8 routing, GQA 64/8 with head dimension 128Source
Input	TextSource
Output	TextSource
Reasoning mode	reasoning and fast-response modes are documentedSource
Tool calling	YesSource

Measurements without matching peer values

These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.

Expert blind preference (Show measurement, test conditions, and source)

Source value: 2.67
Score: 2.67
Metric: points
Unit: points
Category: agentic
Direction: Higher is better
Source type: vendor-reported
Evaluator: Tencent Hy3 model card
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. Vendor internal blind evaluation by 270 experts; 4-point scale; compared against other models in the card.
Context: Internal vendor blind test; do not compare directly with percentage benchmarks.

Source Tencent Hy3 model card

Artificial Analysis Intelligence Index (Show measurement, test conditions, and source)

Source value: 25.7716025731571
Score: 25.7716025731571
Metric: points
Unit: points
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. unknown

Source Artificial Analysis

AA-Briefcase Elo (Show measurement, test conditions, and source)

Source value: 949.65
Score: 949.65
Metric: elo
Unit: elo
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. unknown

Source Artificial Analysis

Output speed (Show measurement, test conditions, and source)

Source value: 94.8680808157153
Score: 94.8680808157153
Metric: tokens/s
Unit: tokens/s
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. unknown
Context: Artificial Analysis performance timing; output speed is decoded after the first streaming chunk. Provider basis: serving-provider median.

Source Artificial Analysis

Time per Intelligence Index Task (Show measurement, test conditions, and source)

Source value: 7.5023408485401895
Score: 7.5023408485401895
Metric: minutes
Unit: minutes
Direction: Lower is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. unknown
Context: Artificial Analysis performance timing; output speed is decoded after the first streaming chunk. Provider basis: serving-provider median.

Source Artificial Analysis

Latency reasoning time (Show measurement, test conditions, and source)

Source value: 21.08190639889799
Score: 21.08190639889799
Metric: seconds
Unit: seconds
Direction: Lower is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. unknown
Context: Artificial Analysis performance timing; output speed is decoded after the first streaming chunk. Provider basis: serving-provider median.

Source Artificial Analysis

Latency input time (Show measurement, test conditions, and source)

Source value: 2.56216244699999
Score: 2.56216244699999
Metric: seconds
Unit: seconds
Direction: Lower is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. unknown
Context: Artificial Analysis performance timing; output speed is decoded after the first streaming chunk. Provider basis: serving-provider median.

Source Artificial Analysis

Answer generation time (end-to-end timing breakdown) (Show measurement, test conditions, and source)

Source value: 5.270476599724497
Score: 5.270476599724497
Metric: seconds
Unit: seconds
Direction: Lower is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. unknown
Context: Artificial Analysis performance timing; output speed is decoded after the first streaming chunk. Provider basis: serving-provider median.

Source Artificial Analysis

End-to-end reasoning time component (Show measurement, test conditions, and source)

Source value: 21.08190639889799
Score: 21.08190639889799
Metric: seconds
Unit: seconds
Direction: Lower is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. unknown
Context: Artificial Analysis performance timing; output speed is decoded after the first streaming chunk. Provider basis: serving-provider median.

Source Artificial Analysis

End-to-end input time component (Show measurement, test conditions, and source)

Source value: 2.56216244699999
Score: 2.56216244699999
Metric: seconds
Unit: seconds
Direction: Lower is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. unknown
Context: Artificial Analysis performance timing; output speed is decoded after the first streaming chunk. Provider basis: serving-provider median.

Source Artificial Analysis

Our tests

Each test shows the latest attempt, with no subsequent fixes to the generated code. Earlier runs remain available in the test details. These results do not contribute to an overall score.

Pelican on a bicycle

A community classic for free-form SVG drawing.

Hy3

The response reached the token limit. No complete recording is available.

Reasoning (requested): high

Current attempt

Run date: 2 Oct 2026
API provider: openrouter
Requested API model: tencent/hy3
API model reported in response: tencent/hy3
Reasoning setting: high
Pinned endpoint: Tencent | tencent/hy3-20260706
Temperature: Provider default
Token limit including reasoning: 8,192 tokens
Test protocol: community-visual-tests-v1
Technical checks: Failed
Response time: 57.5 s

Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.

Earlier attempts (1)

1 Oct 2026API request failedReasoning setting: high

Prompt and test conditions

Generate an SVG of a pelican riding a bicycle

Token limit including reasoning: 8,192 tokens

Task origin (Simon Willison)

Popularity

45.8 / 100

Published prices

Hosting-market price: the median of 5 OpenRouter endpoints from 5 providers (representative: GMICloud), observed on October 7, 2026. Cache prices differ by provider and are not shown.

API input: $0.14 per 1M tokensSource
API output: $0.58 per 1M tokensSource

Predecessor and successor

Sources and data date

Type	Evidence and data date
Speed	Median across serving providers. Output speed after the first streaming chunk, not total response time. Retrieved September 8, 2026. Source: Artificial Analysis · Artificial Analysis (retrieved September 8, 2026)Speed measurement methodology
Price	Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. Source: Pricing · API pricing
Input	Documented modalities. Retrieved September 8, 2026. Source: Model documentation · Model metadata (retrieved September 8, 2026) · Tencent Hy3 model card · Editorial description reviewed October 2, 2026
Output	Documented modalities. Retrieved September 8, 2026. Source: Model documentation · Model metadata (retrieved September 8, 2026) · Tencent Hy3 model card · Editorial description reviewed October 2, 2026
Research status	Research date September 8, 2026. 2 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Not found in the inspected sources	knowledge-cutoff Original wording The exact Hy3 model card does not publish a training-data cutoff.
Additional source	OpenRouter: Popularity index (derived), CC BY 4.0
Additional source	Vercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional source	Creative Commons Attribution 4.0