Extract JSON · 6 checks · max 400 tokens · benched 2026-09-16

Gemini 3.1 Pro Preview vs Gemini 3.8 Flash vs Gemini 3.5 Flash Lite on people mentioned to a json array

Google models side by side on "People mentioned to a JSON array": Gemini 3.5 Flash Lite scores 9/10; Gemini 3.5 Flash Lite is the cheapest answer scoring 8+ at $0.33 per 1,000 runs. Outputs, checks, judge reasons, latency and cost.

The prompt every model received

System

You are a data extraction engine. Reply with JSON only, no prose, no code fences.

User

From the text, return a JSON array of every person mentioned. Each element is an object with keys name, role, email (use null when an email is not given). Preserve the order in which people first appear.

Text: The migration is led by Dana Okafor (Head of Platform, [email protected]). She is supported by Luis Ferreira, a staff engineer, whose address is [email protected]. Questions about billing go to Mei Tanaka, our finance lead; she prefers phone calls and has not shared an email.

Rubric for the judge: A three-element JSON array in order (Dana, Luis, Mei) with correct roles, emails and a JSON null for Mei.

Side by side

Every cell is one OpenRouter call at temperature 0 with the prompt's token cap and reasoning effort "low" where the model supports it. Cost is usage × the catalogue rate in models.json. Quality is one judge call to anthropic/claude-haiku-4.5 against the prompt's rubric, cached per prompt version.

Gemini 3.1 Pro Preview

google/gemini-3.1-pro-preview

1/10

Latency: 2.8s
Cost: $0.00502
Per 1,000: $5.02

133 in · 396 out (380 reasoning) · 5 words · checks 1/6

json_valid
regex
contains
contains
contains
regex

[
  {
    "name": "Dana Okafor",

Judge: Output is incomplete and malformed JSON that cannot be parsed; the task requires a complete three-element array with all required fields.

Gemini 3.8 Flash

google/gemini-3.8-flash

9/10

Latency: 1.5s
Cost: $0.00033
Per 1,000: $0.33

133 in · 61 out · 8 words · checks 6/6

json_valid
regex
contains
contains
contains
regex

[{"name":"Dana Okafor","role":"Head of Platform","email":"[email protected]"},{"name":"Luis Ferreira","role":"staff engineer","email":"[email protected]"},{"name":"Mei Tanaka","role":"finance lead","email":null}]

Judge: Correct structure, order, and content with all three people and proper null for Mei's email, but 'staff engineer' and 'finance lead' lack capitalization consistency with 'Head of Platform'.

Gemini 3.5 Flash Lite

google/gemini-3.5-flash-lite

9/10

top judge scorecheapest scoring 8+

Latency: 566ms
Cost: $0.00032
Per 1,000: $0.33

133 in · 114 out · 33 words · checks 6/6

json_valid
regex
contains
contains
contains
regex

[
  {
    "name": "Dana Okafor",
    "role": "Head of Platform",
    "email": "[email protected]"
  },
  {
    "name": "Luis Ferreira",
    "role": "staff engineer",
    "email": "[email protected]"
  },
  {
    "name": "Mei Tanaka",
    "role": "finance lead",
    "email": null
  }
]

Judge: Correct structure, order, and content with all three people and proper null handling, but 'staff engineer' and 'finance lead' lack capitalization consistency with 'Head of Platform'.

Frequently asked

What does this prompt test?

Extract JSON: A three-element JSON array in order (Dana, Luis, Mei) with correct roles, emails and a JSON null for Mei. The deterministic checks are json_valid, regex, contains, contains, contains, regex.

Which model should I pick for this task?

If the judge's bar of 8/10 is good enough for you, Gemini 3.5 Flash Lite at $0.33 per 1,000 runs. If you need the top score, Gemini 3.5 Flash Lite at $0.33 per 1,000 runs.

Same prompt, other families

Run it yourself

Re-run this prompt on Google with your own OpenRouter key, or tweak the wording and see what changes.