Skip to content

Tencent · Chat

Hy3 (Deep Infra)

Tencent 🇨🇳ChatHy3 familyOpen weights. One of its sources says the weights are publicly available.
From Llmgateway's listing:
Tencent's 295B Mixture-of-Experts model with 21B active parameters, native 256K context support, and three reasoning modes for complex reasoning and agentic tasks.
llminfo IQ
127.2
Market median price
$0.25 blended / 1M
Max context
262K
Listings
38

Evidence and sources

🤗 tencent/Hy3llminfo IQ 127.2, ranked among chat models. From 119 benchmarks. Plenty of data behind it.How llminfo IQ is computedCoding IQ 111.3, 95% interval 105–118. From 13 benchmark families across 3 sources. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.How Coding IQ is computedFair value (52.3). Costs per task about what its benchmark results predict. From 1 benchmark with measured task costs. Little data behind it, so treat it as a rough estimate.How Value for money worksReleased 6 Jul 2026Maker HQ: CN (according to Tencent's own website)
All rankings: #5 in Wikipedia pageviews this week #21 in YouTube mentions this week #43 in Bluesky discussions this week #74 on OpenRouter this week #136 on OpenRouter today #160 on OpenRouter this month #210 on NanoGPT recently #293 on OpenRouter this month #293 on OpenRouter this weekAll rankings: #5 in Wikipedia pageviews this week #21 in YouTube mentions this week #43 in Bluesky discussions this week #74 on OpenRouter this week #136 on OpenRouter today #160 on OpenRouter this month #210 on NanoGPT recently #293 on OpenRouter this month #293 on OpenRouter this weekSources: LMArena data, Models.dev data, AI/ML API data, LLMGateway data, llminfo researchPrimary data from LMArena from Models.dev: name, weights availability from AI/ML API: primary mode from LLMGateway: description from llminfo research: release date

Advantages and trade-offs

Compared with 10 chat models, picked mainly by popularity and IQ

Advantages

Open weights available

The weights are public. 3 of the 7 peers that say either way have open weights.

Trade-offs

We only make a claim when there's enough data on models of the same type. Updated 22 Nov, 04:34 UTC.Prices come from the model's maker where they sell it, otherwise their official partner, otherwise the market median. This comparison uses 7 maker's prices; 3 market median prices.

Input & output

as stated by the model's providers
Text
Text

Capabilities

Supported operations and provider-stated features

ReasoningFunction callingStructured outputPrompt cachingWeb search

Listings

 Orcaroutertencent/hy3-free · freeSource: Models.dev data256K$0.00$0.00——
 NanoGPTtencent/hy3Sources: Models.dev data, NanoGPT262K$0.066cache read $0.029$0.26——
 Eden AIInference by DeepInfradeepinfra/tencent/Hy3Source: Models.dev data262K$0.13cache read $0.033$0.53——
 Hugging FaceInference by DeepInfratencent/Hy3Source: Hugging Face Inference Providers data262K$0.13$0.53——
 Kilotencent/hy3Sources: Kilo AI Gateway data, Models.dev data262K$0.132cache read $0.033$0.528——
 Llmgatewayhy3Source: LLMGateway data262K$0.132cache read $0.033$0.528——
 OpenRouterInference by Tencenttencent/hy3 · fp8Source: OpenRouter data docs262K$0.132cache read $0.033$0.528461,546ms
 Wavespeedtencent/hy3Source: WaveSpeed AI data262K$0.132cache read $0.033$0.528——
 SiliconFlowtencent/Hy3Sources: Models.dev data, SiliconFlow data262K$0.132cache read $0.033$0.528——
 Nous Portaltencent/hy3Source: Nous Portal Inference Gateway data262K$0.14cache read $0.035$0.58——
Show all 34 listings

Showing 10 of 34: the cheapest per 1M input tokens from each provider.

Also sold by Aihubmix, Aimlapi, Atlascloud, Crossmodel, DeepInfra, Flowdot, Gmicloud, Jalapeno, Kenari, Llmgateway Providers, Novita AI, Opencode Go, Qiniu, Redpill, Requesty, Vercel, Zenmux.

4 listings without a price are hidden.

Included in plans

Benchmarks

Named effort levels run from lowest to highest. Select a result to see its source details.

Human preference

Arena ELO47 related benchmarks · 47 results
Source: LMArena data · Nov 2026
BenchmarkScore
Arena ELO
Chinese
Creative Writing
English
Show 43 more benchmarksHide 43 benchmarks
BenchmarkScore
Exclude Ties
Expert
French
German
Hard Prompts
Hard Prompts English
Industry Business and Management and Financial Operations
Industry Entertainment and Sports and Media
Industry Life and Physical and Social Science
Industry Mathematical
Industry Medicine and Healthcare
Industry Software and It Services
Industry Writing and Literature and Language
Japanese
Korean
Longer Query
Multi Turn
Non English
Polish
Russian
Spanish
Style-Controlled
Text Factuality
Text Factuality - Chinese
Text Factuality - Creative Writing
Text Factuality - English
Text Factuality - Exclude Ties
Text Factuality - Expert
Text Factuality - Hard Prompts
Text Factuality - Hard Prompts English
Text Factuality - Industry Business and Management and Financial Operations
Text Factuality - Industry Entertainment and Sports and Media
Text Factuality - Industry Life and Physical and Social Science
Text Factuality - Industry Mathematical
Text Factuality - Industry Medicine and Healthcare
Text Factuality - Industry Software and It Services
Text Factuality - Industry Writing and Literature and Language
Text Factuality - Longer Query
Text Factuality - Multi Turn
Text Factuality - Non English
Text Factuality - Russian
CMT-Benchmark2 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Design Arena EloSource: Design Arena Leaderboards data · Nov 2026

Long context

Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Reasoning

CL-bench · OverallHy3 official Benchmark Appendix · highest reasoning tier unless noted · Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
DeepSearchQA2 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Humanity's Last Exam4 related benchmarks · 4 results
Humanity's Last Exam
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted · text-only · no tools
(no tools, text-only)
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
(with tools, text-only)
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
(with tools)
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted · text-only · with tools
SuperChem2 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Toolathlon VerifiedHy3 official Benchmark Appendix · highest reasoning tier unless noted · Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
WideSearch2 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Agentic

Apex-Agent · Pass@1Hy3 official Benchmark Appendix · highest reasoning tier unless noted · pass@1 · Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Arena AgentSource: LMArena data
BrowseComp2 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
MCP Atlas2 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted · public · Scale April 2026 methodology · 500 tasks
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted · Terminus-2 · XML parser · 4-hour timeout · 500 episodes
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Coding

Arena ELOSource: LMArena data · Nov 2026
CL-benchSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Claw-EvalSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
DeepSWESource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
NL2RepoSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
SkillsBench2 results
Reported by Dots Studio / Xiaohongshu · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Dots Studio official appendix · unstarred comparison cell · version, task count, harness and reasoning setting not independently specified
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Math

ArxivMath2 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
HorizonMath2 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted · pass@12
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Hy-MathHy3 official Benchmark Appendix · highest reasoning tier unless noted · internal · Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated
USAMO 20262 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Science

GPQASource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
GPQA DiamondHy3 official Benchmark Appendix · highest reasoning tier unless noted · Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
PHYBench2 results
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Source: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
Setting not stated

Instruction following

Factuality

Other

Claw-Eval · 20260325 · Pass³Hy3 official Benchmark Appendix · highest reasoning tier unless noted · internal harness · 105 queries · Gemini 3.5 Flash judge · Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
E-BenchHy3 official Benchmark Appendix · highest reasoning tier unless noted · internal · Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy-SWE Max9 related benchmarks · 9 results
Hy-SWE Max
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Hy Expert Blind Evaluation · Average
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted · 270-expert blind evaluation
Hy Internal Evaluation · Commonsense Error Rate
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Hy Internal Evaluation · Multi-turn Issue Rate
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Hy-Backend 2.0
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Hy-CompanyBench
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
Show 3 more resultsHide 3 results
Hy-Euler Pro
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted · with tools
Hy-FinModelBench
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted · internal
Hy-SkillsWorld
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted
NL2Repo-BenchHy3 official Benchmark Appendix · highest reasoning tier unless noted · Claude Code · 250 turns · 12,000s · Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
ProdBench · Pass³Hy3 official Benchmark Appendix · highest reasoning tier unless noted · internal · Pass³ · Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
WildClawBench7 related benchmarks · 7 results
WildClawBench
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
Multimodal (26 tasks)
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · multimodal-input split · 26 tasks
Overall
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · full 60-task suite
Pure-text (34 tasks)
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · pure-text split · 34 tasks
Text-only (35 tasks)
Reported by Tencent Hunyuan · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Hy3 official Benchmark Appendix · highest reasoning tier unless noted · OpenClaw · 35 text-only tasks
Total Cost
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · full 60-task suite
Show 1 more resultHide 1 result
Total Elapsed
Reported by WildClawBench / InternLM · via BenchBoard docs
OpenClaw · full 60-task suite

Data credits: Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · LMArena, CC BY 4.0 (licence) · Baromètre de l’écriture IA, Johannes Eva; public numeric ranking data, © 2026 Johannes Eva, no results licence stated · BullshitBench by Peter Gostev, MIT (licence) · Hugging Face official benchmark leaderboards; licence varies by benchmark · SnakeBench (Greg Kamradt / ARC Prize Foundation), MIT (licence)

Also known as: aihubmix/hy3, deepinfra/tencent/hy3, gmi/tencent/hy3, gmi/tencent/hy3-preview, hy3-free, hy3-preview, hy3-preview-free, novita/hy3, novita/tencent/hy3, opencode_go/hy3, tencent/hy3, tencent/hy3-free, tencent/hy3-preview

Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works