Side by side

See the difference.

Scores, capabilities and costs for the models you’re considering.

Scroll to compare models →

Comparison of Topaz Denoise Image, Step-5-Preview, HunyuanImage 3.5 (Preview)
3 modelstopazTopaz Denoise ImageRemove ×stepfun-aiStep-5-PreviewRemove ×TencentHunyuanImage 3.5 (Preview)Remove ×
What’s differentProfessional photo denoising powered by Topaz Labs. Normal, Strong and Extreme presets clean noise at source resolution; Denoise Max adds generative detail recovery. Best for high-ISO and night photography.600B total and 27B active parameters; accepts images and video, with FP8 and BF16 weights for deployment.
DateAug 11, 2026ListedSep 18, 2026ReleasedSep 22, 2026Released
Listed rateSee provider quotes$1.00 in / $2.70 out/ 1M tokensDeepInfra · all routes ↗See provider quotes
Provider endpoints0 endpoints →6 endpoints →0 endpoints →
AA Intelligence ↗4.3.2Not reported43.73 ↗Effort not specifiedIndependent resultNot reported
AA-Briefcase ↗1.1Not reported1,424.32 ↗Effort not specifiedIndependent resultNot reported
GDPval-AA ↗2.1Not reported1,586.63 ↗Effort not specifiedIndependent resultNot reported
AutomationBench-AA ↗AA Index 4.3.2 collectionNot reported50.96% ↗Effort not specifiedIndependent resultNot reported
Terminal-Bench 4.0 · AA ↗Not reported33.33% ↗Effort not specifiedHarness unreportedIndependent resultNot reported
SciCode ↗AA Index 4.3.2 collectionNot reported58.91% ↗AA · Under reviewEffort not specifiedIndependent resultNot reported
Humanity’s Last Exam ↗AA Index 4.3.2 collectionNot reported46.48% ↗Effort not specifiedIndependent resultNot reported
GDP.pdf ↗AA Index 4.3.2 collectionNot reported14.8% ↗Effort not specifiedIndependent resultNot reported
CritPt ↗AA Index 4.3.2 collectionNot reported20.86% ↗AA · Under reviewEffort not specifiedIndependent resultNot reported
AA-Omniscience ↗AA Index 4.3.2 collectionNot reported16.38 ↗Effort not specifiedIndependent resultNot reported
AA-LCR ↗1.1Not reported88.33% ↗Effort not specifiedIndependent resultNot reported
AA Image ↗2.1Not reportedNot reported969.7 ↗Text-to-image · all categories · (Preview)Independent result95% CI 957.7–981.7 · n=5,586
LiveBench ↗2.0 · 2026-10-14 (dataset 2026-09-02)Not reported62.01% ↗Effort: max · single run · equal mean of 7 suitesIndependent resultNot reported
LiveBench Design ↗2.0 · 2026-10-14 (dataset 2026-09-02)Not reported76.9% ↗Effort: max · single runIndependent resultNot reported
LiveBench Research ↗2.0 · 2026-10-14 (dataset 2026-09-02)Not reported72.8% ↗Effort: max · single runIndependent resultNot reported
LiveBench Browser Use ↗2.0 · 2026-10-14 (dataset 2026-09-02)Not reported24% ↗Effort: max · single runIndependent resultNot reported
LiveBench Coding ↗2.0 · 2026-10-14 (dataset 2026-09-02)Not reported85.6% ↗Effort: max · single runIndependent resultNot reported
LiveBench Agentic ↗2.0 · 2026-10-14 (dataset 2026-09-02)Not reported27.5% ↗Effort: max · single runIndependent resultNot reported
LiveBench Data Analysis ↗2.0 · 2026-10-14 (dataset 2026-09-02)Not reported63.5% ↗Effort: max · single runIndependent resultNot reported
LiveBench Reasoning ↗2.0 · 2026-10-14 (dataset 2026-09-02)Not reported83.8% ↗Effort: max · single runIndependent resultNot reported
Context window—1M—
Max output—64K—
Parameters———
Tool callingNot reportedSupportedNot reported
Structured outputNot reportedSupportedNot reported
Image inputNot reportedSupportedNot reported
Video inputNot reportedSupportedNot reported
ReasoningNot reportedSupportedNot reported
Prompt cachingNot reportedSupportedNot reported
Open weightsNot reportedSupportedNot reported

Add another model

Search all models →