Ctrl K
Back to models

Tencent

Hy3

  • Released
  • Open weights
  • Apache 2.0
Compare
Intelligence Index

Reasoning setting

25.3#43 of 83measured models
Measured · Artificial Analysis v4.3.2
API price from
$0.14 in / $0.58 out / 1M tokOpenRouter · GMICloud · BF16
Context window
256K tok

Performance profile

ReasoningRank #21 of 107
89.7%
GPQA Diamond
CodingRank #36 of 94
48.6%
SciCode
Cost efficiencyRank #10 of 67
352.3pts / $
Intelligence ÷ eval cost

Capabilities

Context window

Modalities

Measured setting

Integration

Reasoning trade-offs

Source (opens in a new tab)
Intelligence Index v4.3.2
Hy3
Cost per task (USD) · log scale
Hy3Reasoning
Intelligence Index
25.3
Cost / task
$0.0718
Evaluation time
8.3m

Pricing & access

API route
OpenRouter · GMICloud · BF16API
Tier & contextStandard

USD per 1M tokens

Input
$0.14
Output
$0.58
Cached input
$0.035

Run locally

GGUF
Context
Estimated memory

~189.1 GiB

RAM + VRAM

Estimate assumptions

One text chat at the selected context, with a 16-bit cache and app allowance. Operating system and image processing are extra; optimized caches can use less.

256 GiB RAMwithout GPU offload

Memory budgetOne text chat
Weights
169.6 GiB
Context cache
2.5 GiB16-bit cache
App allowance
17 GiB
Download
182.2 GB5 files · keep together
Disk space
~206 GB
Available software
Ollama (opens in a new tab)LM Studio (opens in a new tab)Jan (opens in a new tab)KoboldCpp (opens in a new tab)llama.cpp (opens in a new tab)
Get GGUF (opens in a new tab)
Set up in Ollama · Q4_K_M · 8K context

Save the following as Modelfile. The context matches the estimate above.

FROM hf.co/bartowski/Hy3-GGUF:Q4_K_M
PARAMETER num_ctx 8192

Then run these commands in your terminal:

ollama create llmcurrent-local -f Modelfile
ollama run llmcurrent-local

Ollama import instructions (opens in a new tab) · For LM Studio or Jan, open the app’s model search, enter bartowski/Hy3-GGUF, choose Q4_K_M, and set context to 8,192 tokens in the app’s model settings. Software links above open each app’s setup guide.

vLLM setup (opens in a new tab)

Benchmarks

Full leaderboardSource (opens in a new tab)

Humanity’s Last Exam

Frontier reasoning

Hy3 · Reasoning#45 of 126

  1. 1Claude Opus 5.5Max61.4%
  2. 2Claude Fable 5.1Max59.1%
  3. 3Gemini 4 ArgonHigh57.1%
  5. 42Nex-N2-ProReasoning33.7%
  6. 43LongCat 2.0Reasoning33.7%
  7. 44Agnes 2.5 Pro AlphaReasoning33.6%
  8. 45Hy3Reasoning33.5%
  9. 46Inkling SmallReasoning33.3%
  10. 47K2 Horizon 375B A23BReasoning32.0%
  11. 48InklingExtra high31.9%

Peers use their best measured result on this benchmark.

Alternatives

Each model at its highest-scoring measured setting · Intelligence Index v4.3.2 and its evaluation cost per task.