Google

Gemini 4 Neon

Neon matches Carbon's rounded aggregate benchmark score with shorter estimated task time. Start by evaluating high effort for complex coding and analysis; xhigh adds a small overall gain at a larger cost.

google/gemini-4-neonText input and outputFour reasoning levelstext + image inputtext output

Run this model

google/gemini-4-neon

Use an endpoint supported by this model.

Get a key for this setup →
Thinking guide →
const response = await fetch('https://api.minirouter.sh/v1/chat/completions', {  method: 'POST',  headers: {    Authorization: `Bearer ${process.env.MINIROUTER_KEY}`,    'Content-Type': 'application/json',  },  body: JSON.stringify({    model: 'google/gemini-4-neon',    messages: [{ role: 'user', content: 'Why is the sky blue?' }],    max_tokens: 1024,  }),}) const data = await response.json()console.log(data.choices[0].message.content)
Read docs →

Send text, code or extracted document contents. MiniRouter currently exposes text input and output for Neon. A small max_tokens value can still require a large temporary balance hold; see billing for how holds differ from charges.

Model overview

Input / 1M
$3.15
All-in MiniRouter rate
Output / 1M
$15.75
All-in MiniRouter rate
Context
1M
1,000,000 tokens
Max output
1M
tokens
Best fit
Coding and analysis
Repository changes, research and tool workflows.
Reasoning
Low to xhigh
High is Google's default.
Announced
18 Nov 2026
publisher release

Compare text-model alternatives or estimate token costs for your workload.

Independent measurements

Choose how long to think.

High scores 60 on the Artificial Analysis Intelligence Index; xhigh scores 61 and takes about 16 extra seconds to begin its answer. These rounded scores summarize a mix of tasks, so test the extra effort against your own acceptance criteria. Timings use Google's API; your prompt and route will affect latency.

Quality

Intelligence Index · higher is better

0 to 70 points

  • Low44
  • Medium55
  • High60
  • Xhigh61

Wait for an answer

First answer token · lower is better

0 to 60 seconds

  • Low13.93s
  • Medium20.14s
  • High40.64s
  • Xhigh56.51s
Source: Artificial Analysis ↗. Read November 19, 2026. First-party API measurements, not MiniRouter timings.
View chart data
Neon reasoning effort measurements
EffortIntelligence IndexFirst answer token
Low4413.93 s
Medium5520.14 s
High6040.64 s
Xhigh6156.51 s
Choose a model

Give Neon a concrete task and a way to check it.

Start with a failing test, a bounded repository change or a question tied to supplied documents. Judge the finished result and total cost on your own tasks.
  1. [ 01 ]

    Hydrogen for routine work

    Keep simpler extraction, drafting and everyday coding on Hydrogen when its results already meet your needs.

  2. [ 02 ]

    Neon for repeated complex work

    Evaluate Neon on changes that need several tool calls, debugging passes or cross-document reasoning.

  3. [ 03 ]

    Carbon for longer projects

    Google positions Carbon for its hardest, long-horizon work. Neon is a sibling, not a replacement for every Carbon workflow.

Before your first request

Start with text and an explicit effort.

Google's wider model supports multimodal work. The MiniRouter route currently accepts text; use extracted document contents and textual tool results.
  1. [ 01 ]

    Choose low, medium, high or xhigh

    Use reasoning.effort in Chat Completions. Higher effort can spend more time reasoning before the answer begins.

  2. [ 02 ]

    Verify structured results yourself

    The current route does not advertise structured-output parameters. Validate generated JSON in your application.

  3. [ 03 ]

    Leave room for a temporary hold

    A short requested answer does not guarantee a small reservation. Unused reserved credit is released after settlement; the hold is not the final charge.

Route

One route, standard clients.

Authenticate with a MiniRouter bearer key. Caller-provided upstream credentials are not used on this route.
Base URLhttps://api.minirouter.sh/v1
Model IDgoogle/gemini-4-neon
Endpoints/v1/chat/completions/v1/messages/v1/responses
Parametersmax_tokens, temperature, stop, tools, tool_choice, reasoning, include_reasoning
Structured outputNot enforced; a requested format is dropped and the answer is plain text.
ModalitiesInput: text, image · Output: text

Configured providers. Priority follows the live catalog order.

ProviderPriorityInput / 1MOutput / 1MParametersAccess
vercel
1$3.15$15.75max_tokens, temperature, stop, tools, tool_choice, reasoning, include_reasoningMiniRouter balance

Every model, one key.