Roboflow

GoogleGoogleReleased Nov 2026Proprietary

Gemini 4 Carbon

Gemini 4 Carbon scores 89.4% overall on Roboflow's Vision Evals (November 10, 2026 snapshot), #1 of 68 models.

Its best task is counting (90.1%, #1 of 68) and it ranks lowest on identification (93.8%, #10 of 68), a task most models score well on. It is a proprietary model from Google, so commercial use follows Google's terms.

Vision Evals89.4%Field median 64.4% · #1 of 68
Cost$0.0061/ sampleField median $0.0034 · #44 of 68
Speed8.7s/ sampleField median 10.2s · #23 of 68

Gemini 4 Carbon Interactive Demo

Model settings

Thinking level

Max output tokens

Default 65,536 · max 65,536

Sign in to adjust thinking and output length per run.

Results appear here. Add an image or pick an example to run Gemini 4 Carbon.

Gemini 4 Carbon Vision Evals

Vision Evals is Roboflow's ground-truth benchmark: every model runs the same real-world samples across five vision tasks, and answers are scored against ground truth.

Evals updated November 10, 2026Pricing updated November 10, 2026

Strengths and weaknesses

Strengths

TaskScoreMedianRank
Counting90.1%65.5%#1 of 68
Reasoning91.4%59.8%#1 of 68
OCR87.2%64.6%#1 of 52
Object Detection84.5%56.8%#2 of 68

Weak spot

TaskScoreMedianRank
Identification93.8%84.9%#10 of 68

Above the field median, but its lowest rank.

Score by taskScoreRank
Counting
90.1%#1
OCR
87.2%#1
Reasoning
91.4%#1
Object Detection
84.5%#2
Identification
93.8%#10

Gemini 4 CarbonField median

Ranks out of 68 models (52 for OCR)

Results by task

TaskScoreField (0 to 100)RankCost / sampleSpeed
Object Detection (low)
84.5%
±0.3, Mean of 3 runs, range 84.2 to 84.8
#2 of 68$0.01113.83s
Object Detection (high)
85.9%
±0.3, Mean of 3 runs, range 85.6 to 86.2
#1 of 36$0.10284.51s
Counting (low)
90.1%
±1.4, Mean of 3 runs, range 89.2 to 91.9
#1 of 68$0.00246.77s
Counting (high)
86.9%
±2.7, Mean of 3 runs, range 83.8 to 89.2
#3 of 36$0.07570.31s
Identification (low)
93.8%
±0.0, Mean of 3 runs, range 93.8 to 93.8
#10 of 68$0.00246.93s
Identification (high)
96.9%
±0.0, Mean of 3 runs, range 96.9 to 96.9
#3 of 36$0.01920.72s
OCR (low)
87.2%
#1 of 52$0.00537.54s
Single value
80.4%
Transcription
96.6%
Structured JSON
93.0%
Text localization
84.2%
OCR (high)
89.7%
#1 of 52$0.10291.40s
Single value
85.2%
Transcription
99.1%
Structured JSON
93.9%
Text localization
85.2%
Reasoning (low)
91.4%
±1.7, Mean of 3 runs, range 90.1 to 93.4
#1 of 68$0.00256.24s
Reasoning (high)
92.3%
±0.3, Mean of 3 runs, range 92.0 to 92.7
#1 of 55$0.05752.28s
  • Thinking longer helps: 1.4 points higher on object detection at high effort for 8.9x the cost and 6.1x the latency.
  • Thinking longer does not help: 3.2 points lower on counting at high effort for 31.1x the cost and 10.4x the latency.
  • Thinking longer helps: 3.1 points higher on identification at high effort for 7.8x the cost and 3x the latency.
  • Thinking longer helps: 2.5 points higher on ocr at high effort for 19.2x the cost and 12.1x the latency.
  • Thinking longer helps: 0.9 points higher on reasoning at high effort for 22.8x the cost and 8.4x the latency.

Price vs. performance

Score vs. cost

Overall benchmark score against estimated cost per sample, on a log scale. Upper-left is the sweet spot: high quality at low cost.

67 of 68 models plotted (the rest have no measured figure for this axis) · scores and efficiency pooled across all five tasks at low effort · Gemini 4 Carbon highlighted

Gemini 4 Carbon scores are the mean of 3 runs per task at both low and high effort; OCR is a single preliminary run · Methodology

View all Vision Evals →

Gemini 4 Carbon Overview

Gemini 4 Carbon is a multimodal model from Google. It builds on Gemini 4 Argon and supports reasoning over text and images. Google describes improvements in coding and long-running tasks. Model weights are not publicly released.

Deploy Gemini 4 Carbon with an API

Gemini 4 Carbon runs as a hosted REST endpoint through Roboflow Workflows. Pick a task, then hand the prompt to your coding agent or copy the code. Deploying the workflow into a free Roboflow workspace replaces the your-workspace and YOUR_API_KEY placeholders with your own.

Connect your agent to Roboflow (once)

Add the Roboflow MCP server

claude mcp add --transport http roboflow https://mcp.roboflow.com/mcp

Run /mcp and authorize Roboflow in your browser when the OAuth flow opens.

Start a new Claude Code session so the MCP loads, then paste the prompt below (it works the same in any agent).

Deploy this workflow to your Roboflow workspace to use it.

Integrate the Roboflow "Gemini 4 Carbon" workflow into my app.

- Endpoint: POST https://serverless.roboflow.com/<your-workspace>/workflows/gemini-4-carbon-captioning
- Auth: send my Roboflow API key as `api_key` in the request body, read from the ROBOFLOW_API_KEY env var (never hardcode).
- Body: { "api_key": ..., "inputs": { `image`: { type: "url" | "base64", value } } }.
- Billing: this workflow needs no provider API key — inference runs on my Roboflow credits. A BYO provider key can be added to the model step in the Roboflow workflow editor later.

With the Roboflow MCP connected, call `workflows_get` on "gemini-4-carbon-captioning" to read the exact input schema (the source of truth), then `workflows_run` on a sample image to confirm the output shape before writing code (the MCP is authenticated, so this needs no key). Without the MCP, use the contract above.

Before running the app, set up these keys so it does not error at runtime:
- `ROBOFLOW_API_KEY` (sent as `api_key`) from https://app.roboflow.com/settings/api
Create a .gitignore'd .env with these variables, using placeholder values for any I haven't given you. Then pause and tell me directly, in your reply: the full path to the .env file, exactly which keys I need to paste in, and the link to get each one. Wait for me to confirm I've added them before you run anything. Do not run the app until I confirm.

Then add the integration to my codebase: match my project's language, framework, and conventions; read every key from environment variables (never hardcode); add basic error handling; and include a small runnable example. If you can't tell what language my project uses, ask me.
Installpip install inference-sdk

Deploy this workflow to your Roboflow workspace to use it.

# Inference runs on your Roboflow credits — no provider API key needed. To bill your own provider account instead, add an api_key to the model step in the Roboflow workflow editor.
# 1. Import the library
from inference_sdk import InferenceHTTPClient

# 2. Connect to your workflow
client = InferenceHTTPClient(
  api_url="https://serverless.roboflow.com",
  api_key="YOUR_API_KEY"
)

# 3. Run your workflow on an image
result = client.run_workflow(
  workspace_name="your-workspace",
  workflow_id="gemini-4-carbon-captioning",
  images={
    "image": "YOUR_IMAGE.jpg"  # Path to your image file
  },
  use_cache=True  # cache workflow definition for 15 minutes
)

# 4. Get your results
print(result)

Deploy this workflow to your Roboflow workspace to use it.

// Inference runs on your Roboflow credits — no provider API key needed. To bill your own provider account instead, add an api_key to the model step in the Roboflow workflow editor.
const response = await fetch('https://serverless.roboflow.com/your-workspace/workflows/gemini-4-carbon-captioning', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    api_key: 'YOUR_API_KEY',
    inputs: {
      "image": {"type": "url", "value": "IMAGE_URL"}
    }
  })
});

const result = await response.json();
console.log(result);

Deploy this workflow to your Roboflow workspace to use it.

# Inference runs on your Roboflow credits — no provider API key needed. To bill your own provider account instead, add an api_key to the model step in the Roboflow workflow editor.
curl --location 'https://serverless.roboflow.com/your-workspace/workflows/gemini-4-carbon-captioning' \
--header 'Content-Type: application/json' \
--data '{
  "api_key": "YOUR_API_KEY",
  "inputs": {
    "image": {"type": "url", "value": "IMAGE_URL"}
  }
}'

Gemini 4 Carbon License

Proprietary

Gemini 4 Carbon is proprietary: the weights are not distributed, and the Gemini 4 Carbon license is the vendor's commercial terms of service that you accept when you call the API.

Commercial use
Permitted under the vendor terms, typically metered per token or per request, with the vendor usage policy applying to your inputs and outputs.
Modification
Not available. Gemini 4 Carbon weights are closed, so you can configure prompts and use vendor-hosted fine-tuning where it is offered, but you cannot modify the model itself.
Redistribution
Not permitted. You cannot self-host or resell the model; you build on the hosted API instead.

Vendor terms govern data retention, whether your inputs can be trained on, rate limits, and regional availability, and they can change with notice. Review them if you handle regulated or customer data.

Do I need a commercial license for Gemini 4 Carbon?

Proprietary terms are set by the vendor rather than negotiated per project, and no open-source obligation attaches to your code. If you would rather deploy an open model instead of Gemini 4 Carbon, Roboflow's licensing page lists the models whose commercial license is included when you deploy on Roboflow.

Do not hesitate to reach out with questions for your commercial project. Our team will help you start solving business problems on the first call. See Roboflow commercial licensing for the models included in each plan.

Talk to sales

License information is provided as a guide and is not legal advice.

Gemini 4 Carbon Details

Details

Categoryclosed
ModalityMultimodal
Context Window1.0M
LicenseProprietary
Commercial usecommercial use per vendor terms
Release DateNov 2026

Vision Tasks

CaptioningChart Question AnsweringClassificationDocument Question AnsweringImage TaggingMulti-Label ClassificationOCRObject DetectionVision LanguageVisual Question Answering

Usage

Past 30 Days

Frequently Asked Questions About Gemini 4 Carbon Vision

Yes. Gemini 4 Carbon accepts image input and handles OCR, object counting, identification, visual reasoning, and object detection. On Roboflow's Vision Evals its strongest task is Counting at 90.1% (#1 of 68 at low effort). You can test it on your own image in the demo above.

Yes, and it is one of the model's strongest vision skills: it scores 87.2% (#1 of 52 at low effort) on Vision Evals OCR, which covers single readings, full transcriptions, structured JSON, and locating text with boxes. Within that, full transcriptions score 96.6%, and returning the fields of a label or form as JSON scores 93.1%.

Yes. On Vision Evals, Gemini 4 Carbon scores 84.5% mAP@50 on object detection (#2 of 68 at low effort) and 90.1% judge-graded accuracy on object counting.

On our benchmark's task mix, Gemini 4 Carbon averages $0.0061 per sample (#44 of 68 on cost), with an average speed of 8.7s per sample across the benchmark. Actual cost depends on your images and prompts.

On the overall Vision Evals ranking, Gemini 4 Carbon sits #1 of 68 at 89.4%, just ahead of Gemini 4 Argon (87.5%). See the full side-by-side: Gemini 4 Carbon vs Gemini 4 Argon.