DeepSeek V4 Flash vs Grok 4.3: Benchmarks, Pricing & Capabilities Compared
TL;DR — DeepSeek V4 Flash wins for cost · Grok 4.3 wins for general use.
DeepSeek V4 Flash DeepSeek
- Released
- 2026-04-24
- Context window
- 1M tokens
- Input price
- $0.14 / Mtok
- Output price
- $0.28 / Mtok
Grok 4.3 xAI
- Released
- 2026-04-30
- Context window
- 1M tokens
- Input price
- $1.25 / Mtok
- Output price
- $2.50 / Mtok
Benchmark comparison
| Benchmark | DeepSeek V4 Flash | Grok 4.3 |
|---|---|---|
| AA Intelligence Index | 46.5 | 53.2 ✓ |
| Chatbot Arena Elo | 1440 | 1455 ✓ |
| GPQA Diamond | 89.4% | 90.1% ✓ |
| HLE | 32.1% | 35.0% ✓ |
| IF-Bench | 79.2% | 81.3% ✓ |
| LiveCodeBench Reasoning | 63.0% | 64.3% ✓ |
| SciCode | 44.9% | 47.3% ✓ |
| TAU2-bench | 95.0% | 97.7% ✓ |
| TerminalBench-Hard | 35.6% | 37.9% ✓ |
Pricing comparison
| Metric | DeepSeek V4 Flash | Grok 4.3 |
|---|---|---|
| Input ($/Mtok) | $0.14 | $1.25 |
| Output ($/Mtok) | $0.28 | $2.50 |
| Cached input ($/Mtok) | — | — |
| Cost per 1M-token roundtrip (1M in + 1M out) | $0.42 | $3.75 |
Context window & modalities
| Attribute | DeepSeek V4 Flash | Grok 4.3 |
|---|---|---|
| Context window | 1M tokens | 1M tokens |
| Input modalities | text | text, image |
| Output modalities | text | text |
| Knowledge cutoff | — | — |
Verdict by use case
Coding
Insufficient data
Basis: SWE-bench
No shared coding benchmark.
Reasoning
→ Grok 4.3
Basis: GPQA Diamond
DeepSeek V4 Flash 89.4% vs Grok 4.3 90.1% on GPQA Diamond.
Math
Insufficient data
Basis: MATH / AIME
No shared math benchmark.
Long context
Tie
Basis: Context window
DeepSeek V4 Flash 1M tokens vs Grok 4.3 1M tokens.
Cost
→ DeepSeek V4 Flash
Basis: Input $/Mtok
DeepSeek V4 Flash $0.14/Mtok vs Grok 4.3 $1.25/Mtok input.
Changelog & releases
DeepSeek V4 Flash
Released 2026-04-24
Grok 4.3
Released 2026-04-30