Explore AI Chip and System Test Results

See published tests of AI chips and computer systems. Results describe the tested setup; your own equipment may perform differently.

Your AI

Use with your AI

For tasks with several steps, continue in the AI you already use.

CodexClaude CodeCursorWindsurfCline

View instructions

Use my existing Toolmu connection, or follow https://toolmu.com/SKILL.md to set it up. Within my current authorization and budget, call get-hardware-benchmarks-artificial-analysis once with {"queryParams":{}} to retrieve the published datacenter benchmarks. If still processing, observe that same run only; do not retry or make additional Tool requests.

Organize only returned runs by model variant, system and chip count, inference framework/version, precision and execution topology. Keep system throughput, per-request output speed, time to first token and end-to-end metrics separate. Retain supplied units, percentiles and normalization rather than applying chart defaults. Show chip rental prices and passing-concurrency targets only with their returned context. Preserve actual 0, false, null and empty values; mark absent data unknown and report retrieval failures separately. Do not invent chip specifications, a universal winner or production guarantees, and do not launch tests.

Task:
Show the published AI chip and computer system tests from Artificial Analysis. Compare only systems tested in similar conditions. Explain overall system speed separately from how long one AI answer takes, and point out missing details.

Paste into your connected AI's conversation.

Connection guide

Up to 0.036 credits / call

The final charge is based on confirmed usage. Unused reserved credits return to your available balance.

Artificial Analysis

Explore AI Chip and System Test Results

See published tests of AI chips and computer systems. Results describe the tested setup; your own equipment may perform differently.

You receive

Benchmark records may show total system token throughput, output speed for individual requests, time to first token, end-to-end speed and measured power, alongside the tested model, chip count, inference software and precision. Available chip records may include cloud rental prices per GPU-hour. Capacity results may show the highest tested number of concurrent requests meeting the stated speed and first-token latency targets.

Start with your task

Show the published AI chip and computer system tests from Artificial Analysis. Compare only systems tested in similar conditions. Explain overall system speed separately from how long one AI answer takes, and point out missing details.

What to know

These are published test observations, not a test of your own hardware or a production performance guarantee. Compare results with their model, configuration and measurement units intact. Output speed excludes the initial wait for the first token; measured power is not rated chip power. Fields and rental prices may be missing.

Inputs & outputs

Inspect the current inputs, availability, and maximum charge before starting.

Input schema

Input schema
Field / Path	Type	Requirement	Meaning
Query settings`queryParams`	`object`	Optional	Optional object. Leave it empty to retrieve the published datacenter benchmark collections; no search term or identifier is needed.

View raw JSON

{
  "type": "object",
  "required": [],
  "properties": {
    "queryParams": {
      "type": "object",
      "properties": {},
      "additionalProperties": true
    }
  },
  "additionalProperties": false
}

Result schema

Output fields have not been documented yet.

Additional notes

The source chart reports input plus output token throughput, including cache hits, per megawatt of accelerator power. Apply that normalization only when the returned record states it; retain other measurements on their stated basis.

Datacenter benchmark definitions · Endpoint documentation

Questions about this tool.

Inputs, results and details to know before running it.

Get connected

No. A system may handle more requests overall while each answer arrives more slowly. These are different measurements and should stay separate.

Explore AI Chip and System Test Results

Explore AI Chip and System Test Results

You receive

Start with your task

What to know

Inputs & outputs

Input schema

Result schema

Connect your AI assistant

View the published tests

Compare similar systems

Questions about this tool.

You receive

Start with your task

What to know

Inputs & outputs

Input schema

Result schema

Connect your AI assistant

View the published tests

Compare similar systems

Questions about this tool.

Does higher system speed always mean a faster answer?

Is this a complete list of each chip’s capabilities?

Will these tests predict how many requests my computer can handle?