AIEvals

Models

Hunyuan Hy4 Preview

Tencent’s Hunyuan Hy4 Preview, with published Vals AI results.

TencentReleased 28 August 2026Context 1M tokensOutput 64k tokensOpen weights$0.83 in / $2.50 out per Mtok

Family Hunyuan Hy421 benchmarks with a published resultNo first place on the independent columns among the 46 models this site listsRead 17 October 2026

Where it stands

Rank compares tracked models with a score. Published shows the source’s rank and field size, when available.

BenchmarkScoreRankPublishedLeaderBehind
Vals IndexVals AI49.94%26of 3327of 41Gemini 4 Argon18.96 pts
Vibe Code Bench v1.1Vals AI77.48%24of 3825of 106Claude Sonnet 5.514.91 pts
ProgramBenchVals AI0.00%19of 2923of 54Claude Opus 5.518.50 pts
Terminal-Bench 2.1Vals AI55.06%31of 3548of 75Claude Opus 5.532.58 pts
ProofBench v1.1Vals AI75.00%13of 3415of 46Claude Sonnet 5.525.00 pts
LegalBenchVals AI83.76%22of 2746of 149Claude Fable 54.80 pts
Harvey's Legal Agent BenchmarkVals AI9.17%14of 3717of 73Muse Spark 1.216.25 pts
BioMysteryBenchVals AI69.26%11of 1911of 21Claude Sonnet 5.511.85 pts
Code MigrationVals AI47.43%12of 3712of 71Claude Sonnet 5.522.40 pts
CyberBench v1.1Vals AI65.36%21of 3127of 43GPT-6 Sol12.62 pts
EMBVals AI58.08%27of 3734of 69Claude Fable 5.118.59 pts
Finance Agent (v2)Vals AI55.06%15of 3719of 73Gemini 4 Argon10.34 pts
IOIVals AI59.33%15of 3115of 38Gemini 4 Argon40.67 pts
Legal Research BenchVals AI45.19%14of 3714of 72Muse Spark 1.3 Max10.10 pts
MedCodeVals AI43.25%27of 3545of 104Claude Opus 520.32 pts
MedScribeVals AI83.60%29of 3643of 106Claude Opus 5.57.83 pts
Public Benefits Bench v1.1Vals AI68.61%7of 297of 45Claude Opus 58.32 pts
SRE BenchVals AI2.29%13of 1513of 16GPT-6 Astra54.58 pts
Tax Agent BenchVals AI63.71%20of 3723of 64Claude Fable 5.113.93 pts

Nearby on Vals Index

The two models above and below on the Vals Index (Vals AI), on every column where this model has a published number.

ModelVals Index%Vibe Code Bench v1.1%ProgramBench%
GPT-6 LunaOpenAI51.22%81.65%0.50%
Kimi K3Moonshot AI50.70%84.96%2.00%
Hunyuan Hy4 PreviewTencent49.94%77.48%0.00%
Muse Spark 1.2Meta49.29%79.10%–
Qwen3.8 MaxAlibaba48.27%64.70%0.00%
Bold marks a figure better than this model’s on the same column, whichever model it belongs to. Ranks cover the models shown from the stated source.

Results by benchmark

Independent and vendor results retain their own harnesses and effort settings.

BenchmarkIndependent runAs reported by a publisher
Terminal-Bench 4.0Agentic coding
8.08%Vals AI
not published
Terminal-Bench 2.1Agentic coding55.06%Vals AInot published
Vibe Code Bench v1.1Product coding77.48%Vals AInot published
Terminal-Bench Science 0.1Agentic science
1.43%Vals AI
not published
ProgramBenchAgentic coding0.00%Vals AInot published
ProofBench v1.1Mathematics75.00%Vals AInot published
LegalBenchProfessional knowledge83.76%Vals AInot published
Harvey's Legal Agent BenchmarkProfessional knowledge9.17%Vals AInot published
Vals IndexComposite49.94%Vals AInot published
BioMysteryBenchAgentic science69.26%Vals AInot published
Code MigrationAgentic coding47.43%Vals AInot published
CyberBench v1.1Security65.36%Vals AInot published
EMBFinancial modelling58.08%Vals AInot published
Finance Agent (v2)Financial analysis55.06%Vals AInot published
IOICompetitive programming59.33%Vals AInot published
Legal Research BenchProfessional knowledge45.19%Vals AInot published
MedCodeHealthcare administration43.25%Vals AInot published
MedScribeHealthcare administration83.60%Vals AInot published
Public Benefits Bench v1.1Public services68.61%Vals AInot published
SRE BenchSecurity2.29%Vals AInot published
Tax Agent BenchProfessional knowledge63.71%Vals AInot published
Independent column: the suite named under each figure, one harness per suite. Publisher column: each lab’s own table, reproduced on the benchmark’s page. Publisher ranks compare reported scores in the same table row only; ties share a rank and unreported cells are excluded.

Sources

Read on 17 October 2026.