Skip to main content
Radar

Five or fewer confirmed AI changes, with original sources, on mornings when something changed.A free source-linked morning brief.

Start the free Radar Brief
BenchLM
xAI logo
Model A
Grok Realtime

xAI

—

Evidence status unavailable

90% interval unavailable

Grok Realtime vs Hy4 preview

Updated August 28, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

Tencent logo
Model B
Hy4 preview

Tencent

79.16/100

Estimated · Public rank #7

90% interval 69.3–89.0

Decision reading

The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. Use the documented cost, context, and runtime rows instead.

0 results are shared. Category rows based on different benchmark sets are marked directional and do not name a winner.

Which one for your work

Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.

No workload recommendation clears the current evidence threshold.

Use the matched evidence, workload costs, and sourced specifications below instead of treating a point score as a universal answer.

Show secondary and unsupported calls

  • Coding work

    Code generation, repair, and software-engineering tasks

    Not enough matched evidence

    No shared weighted benchmark basis supports a winner.

    Confidence: limited

  • Agentic work

    Tool use, computer use, and multi-step task completion

    Not enough matched evidence

    No shared weighted benchmark basis supports a winner.

    Confidence: limited

  • Long documents

    Prompts that approach the documented context limit

    Not enough matched evidence

    A complete context comparison is not sourced.

    Confidence: limited

  • Chat turn cost

    1K fresh input + 500 output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates

  • Cache-heavy agent loop cost

    200K cached + 20K fresh input + 10K output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates

  • Repository review cost

    50K fresh input + 3K output tokens

    Not enough matched evidence

    A complete comparable API-rate estimate is not available for both models.

    Confidence: listed-rates

What is actually comparable

Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.

Shared results
0
Grok Realtime only
0
Hy4 preview only
28
Like-for-like categories
0 / 8

Category results, on a stated basis

Each row states whether both averages use the same weighted benchmark set. Directional and not-comparable rows remain visible, but they never receive a winner in this template.

Agentic

Not comparable
Grok Realtime
Not measured
Hy4 preview
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Coding

Not comparable
Grok Realtime
Not measured
Hy4 preview
65.7
Weighted basis
0 vs 1 rows
Reading
Not comparable

Reasoning

Not comparable
Grok Realtime
Not measured
Hy4 preview
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Knowledge

Not comparable
Grok Realtime
Not measured
Hy4 preview
60.4
Weighted basis
0 vs 2 rows
Reading
Not comparable

Math

Not comparable
Grok Realtime
Not measured
Hy4 preview
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Multilingual

Not comparable
Grok Realtime
Not measured
Hy4 preview
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Multimodal

Not comparable
Grok Realtime
Not measured
Hy4 preview
66.2
Weighted basis
0 vs 1 rows
Reading
Not comparable

Instruction following

Not comparable
Grok Realtime
Not measured
Hy4 preview
Not measured
Weighted basis
0 vs 0 rows
Reading
Not comparable

Shape of the matched evidence

Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.

A shared-evidence shape is not available.

BenchLM does not draw a radar or infer missing axes when the matched evidence is too sparse.

What each workload costs

Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.

Chat turn

1K fresh input + 500 output tokens

Grok Realtime
API rate not published
Fit state unavailable
Hy4 preview
Self-hosted; infrastructure cost varies
Fits in one request

Grok Realtime has no comparable published API token rate. Hy4 preview has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

Grok Realtime
API rate not published
Fit state unavailable
Hy4 preview
Self-hosted; infrastructure cost varies
Fits in one request

Grok Realtime has no comparable published API token rate. Hy4 preview has no comparable published API token rate.

Cache-heavy agent loop

200K cached + 20K fresh input + 10K output tokens

Grok Realtime
API rate not published
Fit state unavailable
Cached-input rate unavailable
Hy4 preview
Self-hosted; infrastructure cost varies
Fits in one request
Cached-input rate unavailable

Grok Realtime has no comparable published API token rate. Hy4 preview has no comparable published API token rate.

Specification differences

Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.

Documented inputs

Grok Realtime

Not sourced

Hy4 preview

Not sourced

Documented outputs

Grok Realtime

Not sourced

Hy4 preview

Not sourced

Provider availability

Grok Realtime

Not sourced

Hy4 preview

Not sourced

Reasoning profile

Grok Realtime

Non-Reasoning

Hy4 preview

Reasoning

Weight access

Grok Realtime

Proprietary

Hy4 preview

Open Weight

License

Grok Realtime

Proprietary

Hy4 preview

Open Weight

Release date

Grok Realtime

Not sourced

Hy4 preview

2026-08-28

If you already use one of these models
Deployment change
The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
Quality signal
The public evidence has no benchmark result shared by both models, so it does not support a quality verdict.
Workload cost
A complete comparable API-rate estimate is not available for both models.
Context tradeoff
A complete documented context comparison is not available.

Run the same representative tasks against both endpoints before changing production traffic.

Benchmark evidence

The full public result ledger is available for audit without forcing a wide desktop table onto a phone.

Browse raw public benchmark evidence28 rows

Agentic

  • Terminal-Bench 2.1

    Grok Realtime—
    Hy4 preview85.4%
    Source

    Not directly comparable

  • CyberGym

    Grok Realtime—
    Hy4 preview78.4%
    Source

    Not directly comparable

  • WideResearch

    Grok Realtime—
    Hy4 preview83.9%
    Source

    Not directly comparable

  • DRACO

    Grok Realtime—
    Hy4 preview77.2%
    Source

    Not directly comparable

  • MCP Atlas

    Grok Realtime—
    Hy4 preview83.7%
    Source

    Not directly comparable

  • Toolathlon-Verified

    Grok Realtime—
    Hy4 preview74.1%
    Source

    Not directly comparable

  • APEX-Agents

    Grok Realtime—
    Hy4 preview37.1%
    Source

    Not directly comparable

  • skillsBench

    Grok Realtime—
    Hy4 preview62.9%
    Source

    Not directly comparable

  • JobBench

    Grok Realtime—
    Hy4 preview61.7%
    Source

    Not directly comparable

  • Agents' Last Exam

    Grok Realtime—
    Hy4 preview22.8%
    Source

    Not directly comparable

  • AutomationBench

    Grok Realtime—
    Hy4 preview32.1%
    Source

    Not directly comparable

  • BankerToolBench

    Grok Realtime—
    Hy4 preview78.6%
    Source

    Not directly comparable

  • HLE w/ tools

    Grok Realtime—
    Hy4 preview55.4%
    Source

    Not directly comparable

Coding

  • Terminal-Bench 2.1

    Grok Realtime—
    Hy4 preview85.4%
    Source

    Not directly comparable

  • SWE-bench Pro

    Grok Realtime—
    Hy4 preview65.7%
    Source

    Not directly comparable

  • SWE Multilingual

    Grok Realtime—
    Hy4 preview82.9%
    Source

    Not directly comparable

  • deepSwe

    Grok Realtime—
    Hy4 preview64.3%
    Source

    Not directly comparable

  • NL2Repo

    Grok Realtime—
    Hy4 preview58.9%
    Source

    Not directly comparable

  • ProgramBench

    Grok Realtime—
    Hy4 preview17.5%
    Source

    Not directly comparable

  • PostTrain Bench

    Grok Realtime—
    Hy4 preview35.6%
    Source

    Not directly comparable

  • sweMarathon

    Grok Realtime—
    Hy4 preview31.9%
    Source

    Not directly comparable

Reasoning

  • CritPt

    Grok Realtime—
    Hy4 preview16.9%
    Source

    Not directly comparable

Knowledge

  • GPQA

    Grok Realtime—
    Hy4 preview92.3%
    Source

    Not directly comparable

  • GPQA-D

    Grok Realtime—
    Hy4 preview92.3%
    Source

    Not directly comparable

  • HLE

    Grok Realtime—
    Hy4 preview55.4%
    Source

    Not directly comparable

  • HLE w/o tools

    Grok Realtime—
    Hy4 preview43.4%
    Source

    Not directly comparable

Math

  • Apex

    Grok Realtime—
    Hy4 preview74.2%
    Source

    Not directly comparable

Multimodal

  • OfficeQA Pro

    Grok Realtime—
    Hy4 preview66.2%
    Source

    Not directly comparable

Frequently asked questions

Which is better, Grok Realtime or Hy4 preview?

The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. The page therefore keeps the decision tied to the specific documented workload.

Which is better for coding, Grok Realtime or Hy4 preview?

The published evidence does not provide a shared weighted coding basis for both models, so BenchLM does not name a coding winner.

Which is better for agentic tasks, Grok Realtime or Hy4 preview?

The published evidence does not provide a shared weighted agentic tasks basis for both models, so BenchLM does not name a agentic tasks winner.

Which costs less, Grok Realtime or Hy4 preview?

Both models do not have comparable published API token rates, so this page does not name a universal price winner.

Which has the larger context window, Grok Realtime or Hy4 preview?

A complete documented context-window comparison is not available.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated August 28, 2026

Watch Grok Realtime vs Hy4 preview

One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.

Read a sample issue

Join 2,000+ readers.