Tabbit
ResourcesBlogModels
Tabbit LogoTabbit

Tabbit — The AI Browser that Works for You

Topics

  • Best Browsers 2026: Rankings and Top Lists
  • AI Browser Guides and Resources
  • Browser Comparisons: Head-to-Head Guides
  • Browser Reviews and Explainers
  • Browser Alternatives
  • Browser Directory: Guides by Browser
  • Agentic Browser Resources
  • Browser Downloads and Install Guides
  • Browser Productivity Resources
  • SillyTavern Model Guides and Presets

Popular Guides

  • Best Browsers 2026: Top 10 Ranked & Compared
  • Best Free AI Browsers 2026: Plans & Prices Compared
  • Rabbit Browser: What It Is & Why Tabbit Is Better
  • Firefox vs Chrome 2026: Privacy, RAM & Speed Compared
  • Best Lightweight Browsers 2026: 7 Tested for RAM & Speed
  • AI Browser List 2026: 10 Browsers Compared
  • Neobrowser: Norton Neo, Download & Alternatives
  • Edge vs Chrome 2026: Speed, RAM & Battery Tested
  • Best Internet Browser for Android 2026
  • Best AI Browser 2026: Top 9 Tested & Ranked

Events

  • Tabbit Skill Competition
  • KPOP SBTI Fandom Personality Test
  • Tabbit Campus Creator Program
  • fifi's Picks: AI Skills for Research Papers
  • User Survey

About

  • Tabbit Blog
  • Press & Media
English
简体中文English
Review
MediaHy4 Flash

Tencent Hy4 Flash: Plotly Data Analytics Benchmark Results

Original source

Plotly

AuthorPlotly (no individual author identified)

Source date2026-11-16

Tabbit curation2026-11-16

Read original

One-sentence takeaway

Plotly reports that Tencent Hy4 Flash passed 39 of 43 questions (91%, A) in a single run of its DataAnalyticsBench, at a total cost of $0.13 and a median of 29 seconds per question. This is a single-sample result from a benchmark whose questions and dataset have not been made public.

Use cases

  • Tasks this result can help assess: Comparing how models handle analytical questions, ambiguity, impossible requests, data-quality issues, statistical traps, and hidden patterns in Plotly Studio Embedded and its Dash data-analysis environment.

  • Tasks this result should not be generalized to: General-purpose question answering, real enterprise data, non-Plotly toolchains, different prompts, or average performance across repeated runs. The result also does not represent every data-analysis task.

  • Model version covered: The article's Tencent Hy4 flash, released on 2026-11-16; no more specific model snapshot or API version is identified.

  • Test environment or client: Plotly Studio Embedded (a data-analysis agent running in a Dash app), accessed through OpenRouter.

  • Reasoning level and parameters: Reasoning level, temperature, and other sampling parameters are not specified.

Evaluation method

Plotly says it used the same exam, harness, and scoring rules for the new model. The test ran through multiple tool calls and focused on data-analysis queries and chart generation, with less emphasis on text writing. The exam had 43 questions: 20 standard analysis questions, 5 ambiguity questions, 5 impossible requests, 3 data-quality questions, 5 statistical traps, and 5 hidden-pattern questions.

The exam and answer key are private. The dataset is simulated wind-turbine operations data, used to keep answers from appearing in real data sources. Each model was run once. Plotly says repeated tests of other models varied by up to 3 points in a single run, so differences of 1–2 points should be treated as close to a tie.

Key results

  • Hy4 Flash: 39/43 (91%, A), $0.13 for one run, and a median of 29 seconds per question; the full exam took about 33 minutes.

  • Scores by category: standard analysis 20/20, ambiguity 5/5, impossible requests 5/5, data quality 3/3, statistical traps 4/5, and hidden patterns 2/5.

  • Main losses: 3 hidden-pattern questions and 1 statistical-trap question. Plotly checked the transcripts and answer key question by question and says all 4 were genuine misses.

  • Run overhead: about 8 model calls and 4,800 output tokens per question on average; 88% of input tokens came from provider caching; there were no model errors in this run.

  • Plotly's comparison in the same table: Hy4 scored 42/43 at $2.11 and 117 seconds per question; GPT-6 Luna scored 38/43 at $0.13 and 18 seconds per question; Particle-1 scored 40/43 at $0.35 and 16 seconds per question. Score differences between models should be interpreted in light of the single-sample limitation.

Raw data

MetricValue reported by Plotly
Total score39/43 (91%, A)
Standard analysis20/20
Ambiguity questions5/5
Impossible requests5/5
Data quality3/3
Statistical traps4/5
Hidden patterns2/5
Cost for one full run$0.13
Median time per question29 seconds
Full exam timeAbout 33 minutes
Average model calls per questionAbout 8
Output tokens per questionAbout 4,800
Share of input tokens served from cache88%

The article lists OpenRouter rates of about $0.044 per million input tokens and $0.18 per million output tokens. The $0.13 is the cost of this test run and should not be equated directly with per-token rates or with other models tested under different configurations. The article also estimates a cost of about $8 per person per year, assuming 1,000 people, 10 questions per person per day, 250 working days per year, and OpenRouter prices at the time of testing. This is a scenario estimate, not an actual user bill.

Conclusions and limitations

The result supports a limited conclusion: in Plotly's private data-analysis exam and tool environment, Hy4 Flash earned an A in one run at low cost, but scored 2/5 on hidden patterns. Compared with Hy4, it answered 3 fewer questions correctly, had a reported cost of about one-sixteenth as much, and a median time per question of about one-quarter as long.

The exam questions, answers, dataset, full run logs, and sampling parameters have not been made public, so outside readers cannot independently reproduce the 39-question score from the article alone. A single run also cannot establish the variance in model performance; a one- or two-point difference between models should not be treated as a stable advantage. OpenRouter rates and cache hits affect the reported cost, and costs and latency may differ with another provider or execution setup.

Reproduction notes

Reproduction would require Plotly's private DataAnalyticsBench questions and answer key, along with the same version of the Plotly Studio Embedded harness, Dash data environment, OpenRouter access method, and scoring rules. The article says the question bank is shared only with researchers; undisclosed sampling parameters and full execution records remain additional limits on reproducibility. The figures in this note reflect the source article as accessed on 2026-11-17.

Curated by Tabbit

This is a third-party source navigator. Model versions, test environments, and personal experience vary; consult the original source.

Hy4 Flash

Use and compare models in Tabbit

Hy4 Flash

Related reviews

OfficialTencent Hunyuan Research2026-11-16

Tencent Hunyuan Hy4 Flash: Official Benchmarks and WorkBuddy Blind Test

Mediaifanr / APPSO2026-11-16

APPSO’s WorkBuddy Test: Hy4 Flash on a 3D Prototype and Office Tasks

CommunityReddit, r/LocalLLM2026-11-16

Reddit WorkBuddy Ad Metrics Comparison: All Three Models Got the Math Right, Hy4 Flash Had the Lowest Listed Prices

CommunityZhidongxi2026-11-16

Zhidongxi Tests Hy4 Flash: Front-End Prototypes, 3D Creation, and Cross-Tool Blender Tasks

Hy4 Flash

Related showcases & prompts

OfficialOpenRouter

Configuring the Tencent Hy4 Flash API on OpenRouter

OfficialTencent Cloud TokenHub2026-11-16

Tencent Cloud TokenHub: Hy4-Flash Reasoning and API Settings

OfficialOpenClaw documentation

Configure the Tencent Hy4-Flash Provider in OpenClaw

OfficialOpenCode Console

Using Hy4 flash Free in OpenCode Console