Claude Opus 4.7 wins more areas, narrowly.Seed 2.1 Turbo is cheaper and better at images and charts.
Both rank among the ten best models we track in images and charts.
Scores updated · 29 tests both models report · How we compare
Where each one wins
Tests won in each of the five areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software21Claude Opus 4.72 of 3 tests
- FactsGetting facts right instead of making them up10Claude Opus 4.71 of 1 test
- Images and chartsUnderstanding pictures, charts and video14Seed 2.1 Turbo4 of 5 tests
- AgentsCarrying out multi-step tasks on its own22Even2 each
- ReasoningHard problems that need careful thinking11Even1 each
Facts rests on a single test.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
Seed 2.1 Turbo costs 90% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Claude Opus 4.7 pulls ahead
- Short factual questions, answered correctlySimpleQA Verified+22points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+15.5points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+14.5points ahead
Where Seed 2.1 Turbo pulls ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+12.3points ahead
- Visual perception tasks people solve at a glanceBLINK+9points ahead
- Math problems shown in pictures and chartsMathVista+6.1points ahead
Every test, side by side
All 29 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4.7
- Terminal-Bench 2.1Claude Opus 4.7 by 15.583.167.6+15.5
- SWE-bench ProClaude Opus 4.7 by 7.364.357+7.3
- SciCodeSeed 2.1 Turbo by 3.354.557.8+3.3
AgentsEven
- BrowseCompSeed 2.1 Turbo by 5.679.384.9+5.6
- AA ApexAgentsClaude Opus 4.7 by 4.733.929.2+4.7
- MCP AtlasSeed 2.1 Turbo by 377.380.3+3
- OSWorld-VerifiedClaude Opus 4.7 by 1.67876.4+1.6
ReasoningEven
- ARC-AGI-2Claude Opus 4.7 by 14.575.861.3+14.5
- Humanity's Last ExamSeed 2.1 Turbo by 12.342.354.6+12.3
Images and chartsSeed 2.1 Turbo
- BLINKSeed 2.1 Turbo by 970.479.4+9
- CharXiv (RQ)Claude Opus 4.7 by 7.49183.6+7.4
- MathVistaSeed 2.1 Turbo by 6.184.490.5+6.1
- OCRBenchv2Seed 2.1 Turbo by 5.956.962.8+5.9
- MMMU-ProSeed 2.1 Turbo by 3.478.882.2+3.4
Other results14 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- BabyVisionSeed 2.1 Turbo by 40.722.262.9+40.7
- DeepSWEClaude Opus 4.7 by 315423+31
- ERQASeed 2.1 Turbo by 18.852.571.3+18.8
- SimpleVQASeed 2.1 Turbo by 14.656.571.1+14.6
- WorldVQASeed 2.1 Turbo by 12.735.948.6+12.7
- RealWorldQASeed 2.1 Turbo by 10.775.686.3+10.7
- ToolathlonClaude Opus 4.7 by 10.259.349.1+10.2
- OfficeQA ProClaude Opus 4.7 by 9.580.671.1+9.5
- Finance Agent v1.1Claude Opus 4.7 by 8.464.456+8.4
- CyberGymClaude Opus 4.7 by 6.173.167+6.1
- ChartQAProSeed 2.1 Turbo by 5.465.570.9+5.4
- EmbSpatial-BenchSeed 2.1 Turbo by 5.377.282.5+5.3
- DynaMathSeed 2.1 Turbo by 563.168.1+5
- DUDEClaude Opus 4.7 by 1.484.583.1+1.4
Questions people ask
Which is better, Claude Opus 4.7 or Seed 2.1 Turbo?
Claude Opus 4.7 wins two of the five areas where both have results: coding and facts. Seed 2.1 Turbo wins images and charts, and costs 90% less. They are level on agents and reasoning.
Which is better for coding?
Claude Opus 4.7. It wins 2 of the 3 coding tests both models report; Seed 2.1 Turbo wins 1.
Which is cheaper?
Claude Opus 4.7 costs $5.00 per million input tokens and $25.00 per million output tokens; Seed 2.1 Turbo costs $0.50 and $2.50. That makes Seed 2.1 Turbo about 90% cheaper for the same work.
How do you compare the two?
We use the 29 benchmark tests both models have published scores on. The verdict counts the 15 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 14 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.