Tencentopen-weight
Released: 2026-07-06Artificial AnalysisHy3
Hy3 is a 295B-parameter Mixture-of-Experts model from Tencent (21B active, 192 experts with top-8 routing) built for reasoning, agentic workflows, and real-world production use. It supports a configurable reasoning effort:...
Intelligence Index
—/ 100
Artificial Analysis Intelligence Index (independent composite)Throughput Speed
76tok/sec
Output streaming throughputTime to First Answer
48s
Time to first answer token, including any thinkingCost to Run the Index
—USD
Artificial Analysis cost of the full Index runMeasured results · Artificial Analysis
Independent evaluations of Hy3. Values are reported as published; nothing here is estimated or derived. Snapshot 2026-11-15.
- Intelligence Index
- —
- Cost to run the Index
- —
- Time per task
- —
- Omniscience index
- —
- Reasoning model
- Yes
- Cached-input price
- $0.034 / 1M
Reasoning & knowledge
- CritPt4.9%
- GPQA Diamond89.7%
- HLE-Diamond11.6%
- Humanity's Last Exam33.5%
Coding & terminal agents
- SciCode48.6%
- Terminal-Bench 2.164.4%
- Terminal-Bench 4.00.5%
Agentic work
- AA-Briefcase v1.1 (Elo)948 Elo
- AutomationBench-AA18.4%
- GDPval-AA (normalized Elo)1069 Elo
- τ³-Banking22.9%
Long context & documents
- AA-LCR (long-context reasoning)79.0%
- GDP.pdf11.2%
Headline Evaluation Breakdown
Six independently measured evaluations (Artificial Analysis)Coding & agents · Terminal-Bench 4.00.5%
Long context · AA-LCR79.0%
Knowledge · Humanity's Last Exam33.5%
Sourced Benchmark Evaluations (11)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| GPQA | Nearing Saturation | 89.7% | 78 | 2026-11-15 | independent | Source ↗ |
| Humanity's Last Exam | Active | 33.5% | 93 | 2026-11-15 | independent | Source ↗ |
| SciCode | Active | 48.6% | 93 | 2026-11-15 | independent | Source ↗ |
| τ³-Banking | Active | 22.9% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench 2.1 | Active | 64.4% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench 4.0 | Active | 0.5% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-LCR (Long Context Reasoning) | Active | 79% | 93 | 2026-11-15 | independent | Source ↗ |
| CritPt | Active | 4.9% | 93 | 2026-11-15 | independent | Source ↗ |
| HLE-Diamond | Active | 11.6% | 93 | 2026-11-15 | independent | Source ↗ |
| GDP.pdf | Active | 11.2% | 93 | 2026-11-15 | independent | Source ↗ |
| AutomationBench-AA | Active | 18.4% | 93 | 2026-11-15 | independent | Source ↗ |
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run tencent-hy3
# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{
"model": "tencent-hy3",
"messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
}'Evaluation & Sourcing Notes
Scores listed for Hy3 represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.