- Chart data
- Anthropic
- Claude 4 benchmarks
Claude 4 benchmarks
The numbers behind a chart Anthropic published, redrawn natively. Check the original, study the data, then make your own.
Simplified redraw
Drawn by us from the numbers below. Hover, tap or use the arrow keys.A simplified redraw: it shows the numbers in our own look, not the look of the original. One accent marks what the chart is about. Values marked with ~ in the table are estimates.
The numbers
First 10 of 65 points are free| Group | Series | Value | Source | Conf. |
|---|---|---|---|---|
| Agentic coding SWE-bench Verified | Claude Opus 4 | 72.5% | Estimate | 0.97 |
| Agentic coding SWE-bench Verified | Claude Sonnet 4 | 72.7% | Estimate | 0.97 |
| Agentic coding SWE-bench Verified | Claude Sonnet 3.7 | 62.3% | Estimate | 0.97 |
| Agentic coding SWE-bench Verified | OpenAI o3 | 69.1% | Estimate | 0.97 |
| Agentic coding SWE-bench Verified | OpenAI GPT-4.1 | 54.6% | Estimate | 0.97 |
| Agentic coding SWE-bench Verified | Gemini 2.5 Pro Preview (US-06) | 63.2% | Estimate | 0.97 |
| Agentic coding SWE-bench Verified (with parallel test-time compute) | Claude Opus 4 | 79.4% | Estimate | 0.97 |
| Agentic coding SWE-bench Verified (with parallel test-time compute) | Claude Sonnet 4 | 80.2% | Estimate | 0.97 |
| Agentic coding SWE-bench Verified (with parallel test-time compute) | Claude Sonnet 3.7 | 70.3% | Estimate | 0.97 |
| Agentic terminal coding Terminal-bench | Claude Opus 4 | 43.2% | Estimate | 0.97 |
Max data, free with your email for now
See all 65 points and download the files
The full table, the JSON, a CSV and the agent skill file for this chart. These are Max in the plan; while this is a prototype your email unlocks them.
Unlocks all 35 giants. One email a week, leave any time. What you get · Log in
| Agentic terminal coding Terminal-bench | Claude Sonnet 4 | 35.5% | Estimate | 0.97 |
|---|---|---|---|---|
| Agentic terminal coding Terminal-bench | Claude Sonnet 3.7 | 35.2% | Estimate | 0.97 |
| Agentic terminal coding Terminal-bench | OpenAI o3 | 30.2% | Estimate | 0.97 |
| Agentic terminal coding Terminal-bench | OpenAI GPT-4.1 | 30.3% | Estimate | 0.30 |
| Agentic terminal coding Terminal-bench (with parallel test-time compute) | Claude Opus 4 | 50.2% | Estimate | 0.30 |
| Agentic terminal coding Terminal-bench (with parallel test-time compute) | Claude Sonnet 4 | 41.3% | Estimate | 0.30 |
| Graduate-level reasoning GPQA Diamond | Claude Opus 4 | 79.6% | Estimate | 0.30 |
| Graduate-level reasoning GPQA Diamond | Claude Sonnet 4 | 75.4% | Estimate | 0.30 |
| Graduate-level reasoning GPQA Diamond | Claude Sonnet 3.7 | 78.2% | Estimate | 0.97 |
| Graduate-level reasoning GPQA Diamond | OpenAI o3 | 83.3% | Estimate | 0.97 |
| Graduate-level reasoning GPQA Diamond | OpenAI GPT-4.1 | 66.3% | Estimate | 0.30 |
| Graduate-level reasoning GPQA Diamond | Gemini 2.5 Pro Preview (US-06) | 83% | Estimate | 0.97 |
| Graduate-level reasoning GPQA Diamond (with parallel test-time compute) | Claude Opus 4 | 83.3% | Estimate | 0.50 |
| Graduate-level reasoning GPQA Diamond (with parallel test-time compute) | Claude Sonnet 4 | 83.8% | Estimate | 0.50 |
| Agentic tool use TAU-bench Retail | Claude Opus 4 | 81.4% | Estimate | 0.97 |
| Agentic tool use TAU-bench Retail | Claude Sonnet 4 | 80.5% | Estimate | 0.97 |
| Agentic tool use TAU-bench Retail | Claude Sonnet 3.7 | 81.2% | Estimate | 0.97 |
| Agentic tool use TAU-bench Retail | OpenAI o3 | 70.4% | Estimate | 0.97 |
| Agentic tool use TAU-bench Retail | OpenAI GPT-4.1 | 68% | Estimate | 0.97 |
| Agentic tool use TAU-bench Airline | Claude Opus 4 | 59.6% | Estimate | 0.97 |
| Agentic tool use TAU-bench Airline | Claude Sonnet 4 | 60% | Estimate | 0.97 |
| Agentic tool use TAU-bench Airline | Claude Sonnet 3.7 | 58.4% | Estimate | 0.97 |
| Agentic tool use TAU-bench Airline | OpenAI o3 | 52% | Estimate | 0.97 |
| Agentic tool use TAU-bench Airline | OpenAI GPT-4.1 | 49.4% | Estimate | 0.97 |
| Multilingual Q&A MMMLU | Claude Opus 4 | 88.8% | Estimate | 0.97 |
| Multilingual Q&A MMMLU | Claude Sonnet 4 | 86.5% | Estimate | 0.97 |
| Multilingual Q&A MMMLU | Claude Sonnet 3.7 | 85.9% | Estimate | 0.97 |
| Multilingual Q&A MMMLU | OpenAI o3 | 88.8% | Estimate | 0.97 |
| Multilingual Q&A MMMLU | OpenAI GPT-4.1 | 83.7% | Estimate | 0.97 |
| Visual reasoning MMMU (validation) | Claude Opus 4 | 76.5% | Estimate | 0.50 |
| Visual reasoning MMMU (validation) | Claude Sonnet 4 | 74.4% | Estimate | 0.50 |
| Visual reasoning MMMU (validation) | Claude Sonnet 3.7 | 75% | Estimate | 0.50 |
| Visual reasoning MMMU (validation) | OpenAI o3 | 82.9% | Estimate | 0.50 |
| Visual reasoning MMMU (validation) | OpenAI GPT-4.1 | 74.8% | Estimate | 0.50 |
| Visual reasoning MMMU (validation) | Gemini 2.5 Pro Preview (US-06) | 79.6% | Estimate | 0.50 |
| High school math competition AIME 2025 | Claude Opus 4 | 75.5% | Estimate | 0.97 |
| High school math competition AIME 2025 | Claude Sonnet 4 | 70.5% | Estimate | 0.97 |
| High school math competition AIME 2025 | Claude Sonnet 3.7 | 54.8% | Estimate | 0.97 |
| High school math competition AIME 2025 | OpenAI o3 | 88.9% | Estimate | 0.97 |
| High school math competition AIME 2025 | Gemini 2.5 Pro Preview (US-06) | 83% | Estimate | 0.97 |
| High school math competition AIME 2025 (with parallel test-time compute) | Claude Opus 4 | 90% | Estimate | 0.97 |
| High school math competition AIME 2025 (with parallel test-time compute) | Claude Sonnet 4 | 85% | Estimate | 0.97 |
| Graduate-level reasoning Terminal-bench | Claude Opus 4 | 79.6% | Estimate | 0.50 |
| Visual reasoning MMMLU (validation) | Claude Opus 4 | 76.5% | Estimate | 0.50 |
| Graduate-level reasoning Terminal-bench | Claude Sonnet 4 | 75.4% | Estimate | 0.50 |
| Visual reasoning MMMLU (validation) | Claude Sonnet 4 | 74.4% | Estimate | 0.50 |
| Graduate-level reasoning Terminal-bench | Claude Sonnet 3.7 | 35.2% | Estimate | 0.50 |
| Visual reasoning MMMLU (validation) | Claude Sonnet 3.7 | 75% | Estimate | 0.50 |
| Graduate-level reasoning Terminal-bench | OpenAI o3 | 30.2% | Estimate | 0.50 |
| Visual reasoning MMMLU (validation) | OpenAI o3 | 82.9% | Estimate | 0.50 |
| Graduate-level reasoning Terminal-bench | OpenAI GPT-4.1 | 66.3% | Estimate | 0.50 |
| Visual reasoning MMMLU (validation) | OpenAI GPT-4.1 | 74.8% | Estimate | 0.50 |
| Agentic terminal coding (pass@1) | Gemini 2.5 Pro Preview (US-06) | 83% | Estimate | 0.50 |
| Graduate-level reasoning Terminal-bench | Gemini 2.5 Pro Preview (US-06) | 83% | Estimate | 0.50 |
| Visual reasoning MMMLU (validation) | Gemini 2.5 Pro Preview (US-06) | 79.6% | Estimate | 0.50 |
Max subscribers get the same data by API. API docs
API · Max
Use this chart via API
On the Max plan you can pull this chart's numbers with the provenance of each one, as JSON or CSV, with its redraw spec, or call it from an agent over MCP. Chart id b773ef9530ef.
| Path | Returns | Plan | |
|---|---|---|---|
| GET | /api/v1/charts/b773ef9530ef | Title, publisher, date, form, point count; with a Max key, the data link | Open |
| GET | /api/v1/charts/b773ef9530ef/data | Every point with its provenance and confidence, plus the redraw spec. Add ?format=csv for CSV | Max |
| GET | /api/v1/charts/b773ef9530ef/render?view=1 | The draw-only view this page uses | Open |
| POST | /mcp | Tool get_chart_data with this id, for Claude, Cursor and other agents | Max |
curl -s https://viral-chart.com/api/v1/charts/b773ef9530ef/data \
-H "Authorization: Bearer $VIRALCHART_KEY"claude mcp add --transport http viralchart https://viral-chart.com/mcp \
--header "Authorization: Bearer $VIRALCHART_KEY"60 requests a minute per key; over that you get 429 with Retry-After. Up to 5 keys per account, made on your account page and shown once. Without Max, the data route answers 402 with a link to the plans. The numbers describe a chart its publisher owns, and every file carries a license note.
Make one like this
This chart's shape, your numbers.
Live preview. Everything stays on your device.
Pick the form
A simplified form: our own look, not the look of the original. Your numbers start from this chart's data.
Swap in your numbers
These are this chart's own names and numbers so you can see the shape. Type over any name or number; the preview redraws as you go. Replace every one with yours before you post.
These are this chart's numbers so you can see the shape. Replace every one with yours before you post.
Your brand
Export the picture
Free pictures carry a small "Made with ViralChart" line in the corner. That line is how other people find us.
More data from Anthropic
All chart dataQuestions people ask
How accurate are these numbers?
They are estimated from the image by two independent passes, with confidence 0.76 on a 0 to 1 scale. Printed numbers read well; values read off an axis can be a few percent off.
Can I reuse this data?
The numbers describe a chart that Anthropic published. The image, layout and styling belong to them. Use the numbers with a credit to the publisher, and make your picture from your own data.
How do I get the file?
The JSON, CSV and agent skill file are under the email gate on this page. In this prototype they are static files at /api/charts/b773ef9530ef.json and /api/charts/b773ef9530ef.skill.md.