Claude Opus 4.6
VS
DeepSeek V3.1 Terminus
Claude Opus 4.6 vs DeepSeek V3.1 Terminus: which wins at real work?
22 task areas · same graded test runs · rank comparison only, so 0–100 and Elo collections never mix raw scores.
Claude Opus 4.6 wins 16 of 22 task areas we tested; DeepSeek V3.1 Terminus takes 6. DeepSeek V3.1 Terminus costs 24.6× less per token ($1.22 vs $30 per 1M).
DeepSeek V3.1 Terminus costs 24.6× less per token ($1.22 vs $30 per 1M).
Task by task
| Task area | Claude Opus 4.6 | DeepSeek V3.1 Terminus | Winner |
|---|---|---|---|
| Translation & Localization |
#2
/ 110
Excellent
|
#101
/ 110
Strong
|
Claude Opus 4.6 |
| AI Strategy |
#5
/ 126
Strong
|
#101
/ 126
Usable
|
Claude Opus 4.6 |
| Sales |
#4
/ 110
Strong
|
#97
/ 110
Needs editing
|
Claude Opus 4.6 |
| Legal & HR |
#8
/ 110
Excellent
|
#95
/ 110
Strong
|
Claude Opus 4.6 |
| Content & Brand |
#14
/ 124
Strong
|
#100
/ 124
Usable
|
Claude Opus 4.6 |
| Chef / Home Cooking |
#12
/ 126
Strong
|
#95
/ 126
Usable
|
Claude Opus 4.6 |
| Creative & Comedy | #10 / 110 | #89 / 110 | Claude Opus 4.6 |
| Coding |
#18
/ 115
Excellent
|
#92
/ 115
Usable
|
Claude Opus 4.6 |
| Executive Assistant |
#18
/ 112
Strong
|
#77
/ 112
Usable
|
Claude Opus 4.6 |
| Investor & Pitch |
#8
/ 66
Strong
|
#64
/ 66
Needs editing
|
Claude Opus 4.6 |
| Data & Analytics |
#10
/ 110
Excellent
|
#56
/ 110
Excellent
|
Claude Opus 4.6 |
| Presentations & Decks |
#54
/ 110
Excellent
|
#91
/ 110
Strong
|
Claude Opus 4.6 |
| Structured Output |
#98
/ 113
Strong
|
#63
/ 113
Excellent
|
DeepSeek V3.1 Terminus |
| Training & Education |
#1
/ 110
Excellent
|
#31
/ 110
Excellent
|
Claude Opus 4.6 |
| RAG, Safety & Grounding |
#59
/ 113
Excellent
|
#85
/ 113
Excellent
|
Claude Opus 4.6 |
| Landing Pages |
#23
/ 72
Strong
|
#44
/ 72
Strong
|
Claude Opus 4.6 |
| Frontend & Landing Pages |
#51
/ 109
Needs editing
|
#33
/ 109
Needs editing
|
DeepSeek V3.1 Terminus |
| Research & Competitive Analysis |
#29
/ 110
Strong
|
#46
/ 110
Strong
|
Claude Opus 4.6 |
| Knowledge & Docs |
#66
/ 110
Usable
|
#50
/ 110
Usable
|
DeepSeek V3.1 Terminus |
| Customer Support |
#81
/ 113
Usable
|
#69
/ 113
Usable
|
DeepSeek V3.1 Terminus |
| Product & Project Management |
#45
/ 110
Strong
|
#36
/ 110
Strong
|
DeepSeek V3.1 Terminus |
| Summarization & Meeting Notes |
#34
/ 110
Excellent
|
#32
/ 110
Excellent
|
DeepSeek V3.1 Terminus |
Rank = position among every model config we tested in that task area (lower is better). Sorted by biggest gap first.
Same task, both models — judged
Both models answered the same test case; an independent judge graded each. Quotes are the judge's actual rationale.
Coding
Mutable default argument (Bug Fixing)
“The model perfectly diagnoses and fixes the root cause of the bug. It uses the standard Python idiom (`items=None`) to avoid the mutable default argument trap, keeps the changes minimal, and provides a clear, accurate explanation of why the original code failed and why the fix works.”
“The model correctly identifies the root cause (mutable default arguments) and provides a perfect primary fix with a great explanation. However, it loses significant points on correctness and minimality by volunteering alternative solutions that are flawed. The sentinel object alternative demonstrates a misunderstanding of Python object identity and will crash, while the conditional assignment alternative subtly changes the function's behavior for explicitly passed empty lists.”
Frequently asked
Is Claude Opus 4.6 better than DeepSeek V3.1 Terminus?
Across 22 task areas we benchmarked, Claude Opus 4.6 ranks higher in 16 and DeepSeek V3.1 Terminus in 6.
Which is cheaper, Claude Opus 4.6 or DeepSeek V3.1 Terminus?
DeepSeek V3.1 Terminus costs 24.6× less per token ($1.22 vs $30 per 1M).
What is Claude Opus 4.6 better at?
Claude Opus 4.6 out-ranks DeepSeek V3.1 Terminus at Translation & Localization, AI Strategy, Sales.
What is DeepSeek V3.1 Terminus better at?
DeepSeek V3.1 Terminus out-ranks Claude Opus 4.6 at Structured Output, Frontend & Landing Pages, Knowledge & Docs.
More comparisons
This page is Spring Prompt, running in public
We just did this for every model. Do it for your prompt.
The rankings above come from running real tasks through real models and scoring every output. Spring Prompt is that same engine — pointed at your prompt, your test cases, and your definition of good.
- Generate test cases from your prompt — no eval set required to start.
- Compare models side by side with quality, cost and latency in one matrix.
- Optimise the winner until the scores say it's ready to ship.
Prompt × model results
12 test cases · 3 evals