Compare / head-to-head
DeepSeek V4 ProvsGrok 4.20 Reasoning
DeepSeek V4 Pro leads 4 of 5 shared benchmarks. DeepSeek V4 Pro is 2.9x 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
4 – 1
DeepSeek V4 Pro leads
Cheaper per token
DeepSeek V4 Pro
2.9x cheaper, input + output
Larger context
Tie
both 1M tokens
Provider uptime (30d)
100% · —
DeepSeek · xAI
DeepSeek V4 Pro
DeepSeek · released 2026-08-13
text
Grok 4.20 Reasoning
xAI · released 2026-03-09
textvision
Quality
Benchmark matrix
5 shared · 6 only DeepSeek V4 Pro · 0 only Grok 4.20 Reasoning| Benchmark | DeepSeek V4 Pro | Grok 4.20 Reasoning | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 5 | ||||
| GPQA Diamond | 90.9% ↗epoch | 89.3% ↗epoch | +1.6 pt | DeepSeek V4 Pro |
| FrontierMath Tier 4 v2 (Epoch AI run) | 2.4% ↗epoch | 17.1% ↗epoch | -14.7 pt | Grok 4.20 Reasoning |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 45.3% ↗epoch | 44.9% ↗epoch | +0.4 pt | DeepSeek V4 Pro |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 96.7% ↗epoch | 92.2% ↗epoch | +4.5 pt | DeepSeek V4 Pro |
| SimpleQA Verified | 47% ↗epoch | 30.2% ↗epoch | +16.8 pt | DeepSeek V4 Pro |
| Only DeepSeek V4 Pro reports · 6 | ||||
| AIME 2026 | 96.7% ↗matharena ⚠ | not reported | — | — |
| LMArena Elo | 1450.6 ↗ | not reported | — | — |
| DeepSWE v1.1 | 62.7% ↗ | not reported | — | — |
| Humanity's Last Exam (with tools) | 60% ↗ | not reported | — | — |
| SWE-bench Verified (Epoch AI run) | 77.6% ↗epoch | not reported | — | — |
| Terminal-Bench 2.1 | 87.9% ↗ | not reported | — | — |
| Only Grok 4.20 Reasoning reports · 0 | ||||
| Grok 4.20 Reasoning reports nothing DeepSeek V4 Pro does not. | ||||
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 DeepSeek V4 Pro minus Grok 4.20 Reasoning 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 | DeepSeek V4 Pro | Grok 4.20 Reasoning | Edge |
|---|---|---|---|
| Context window | 1M tokens | 1M tokens | Tie |
| Max output | 384K tokens | — | — |
| Input price / 1M | $0.435 ↗ | $1.25 ↗ | DeepSeek V4 Pro |
| Output price / 1M | $0.87 ↗ | $2.5 ↗ | DeepSeek V4 Pro |
| Cached input / 1M | $0.004 ↗ | $0.2 ↗ | DeepSeek V4 Pro |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $1.31 | $3.75 | DeepSeek V4 Pro |
| Modalities | text | text · vision | Grok 4.20 Reasoning |
| Released | 2026-08-13 | 2026-03-09 | — |
| Cited benchmark scores | 11 | 5 | — |
Reliability
Provider status
Last 60 days · refreshed daily on this page · live on /statusMore matchups
DeepSeek V4 Pro vs …
Models sharing the most benchmarksMore matchups
Grok 4.20 Reasoning vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run DeepSeek V4 Pro and Grok 4.20 Reasoning on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.