Compare / head-to-head
Claude Sonnet 5vsDeepSeek V4 Pro
DeepSeek V4 Pro leads 4 of 6 shared benchmarks. DeepSeek V4 Pro is 9.2x cheaper per token. Both offer a 1M-token context window.
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
2 – 4
DeepSeek V4 Pro leads
Cheaper per token
DeepSeek V4 Pro
9.2x cheaper, input + output
Larger context
Tie
both 1M tokens
Provider uptime (30d)
100% · 100%
Anthropic · DeepSeek
Claude Sonnet 5
Anthropic · released 2026-06-30
textvision
DeepSeek V4 Pro
DeepSeek · released 2026-08-13
text
Quality
Benchmark matrix
6 shared · 2 only Claude Sonnet 5 · 5 only DeepSeek V4 Pro| Benchmark | Claude Sonnet 5 | DeepSeek V4 Pro | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 6 | ||||
| GPQA Diamond | 90.5% ↗epoch | 90.9% ↗epoch | -0.4 pt | DeepSeek V4 Pro |
| FrontierMath Tier 4 v2 (Epoch AI run) | 29.3% ↗epoch | 2.4% ↗epoch | +26.9 pt | Claude Sonnet 5 |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 65.6% ↗epoch | 45.3% ↗epoch | +20.3 pt | Claude Sonnet 5 |
| Humanity's Last Exam (with tools) | 57.4% ↗ | 60% ↗ | -2.6 pt | DeepSeek V4 Pro |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 94.7% ↗epoch | 96.7% ↗epoch | -2 pt | DeepSeek V4 Pro |
| SimpleQA Verified | 33.7% ↗epoch | 47% ↗epoch | -13.3 pt | DeepSeek V4 Pro |
| Only Claude Sonnet 5 reports · 2 | ||||
| SWE-bench Verified | 85.2% ↗ | not reported | — | — |
| Humanity's Last Exam (no tools) | 43.2% ↗ | not reported | — | — |
| Only DeepSeek V4 Pro reports · 5 | ||||
| AIME 2026 | not reported | 96.7% ↗matharena ⚠ | — | — |
| LMArena Elo | not reported | 1450.6 ↗ | — | — |
| DeepSWE v1.1 | not reported | 62.7% ↗ | — | — |
| SWE-bench Verified (Epoch AI run) | not reported | 77.6% ↗epoch | — | — |
| Terminal-Bench 2.1 | not reported | 87.9% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is Claude Sonnet 5 minus DeepSeek V4 Pro in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | Claude Sonnet 5 | DeepSeek V4 Pro | Edge |
|---|---|---|---|
| Context window | 1M tokens | 1M tokens | Tie |
| Max output | 128K tokens | 384K tokens | DeepSeek V4 Pro |
| Input price / 1M | $2 ↗ | $0.435 ↗ | DeepSeek V4 Pro |
| Output price / 1M | $10 ↗ | $0.87 ↗ | DeepSeek V4 Pro |
| Cached input / 1M | $0.2 ↗ | $0.004 ↗ | DeepSeek V4 Pro |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $12 | $1.31 | DeepSeek V4 Pro |
| Modalities | text · vision | text | Claude Sonnet 5 |
| Released | 2026-06-30 | 2026-08-13 | — |
| Cited benchmark scores | 8 | 11 | — |
Reliability
Provider status
Last 60 days · refreshed daily on this page · live on /statusMore matchups
Claude Sonnet 5 vs …
Models sharing the most benchmarksMore matchups
DeepSeek V4 Pro vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run Claude Sonnet 5 and DeepSeek V4 Pro on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.