token.app › Compare › Claude 3 Haiku vs DeepSeek V4 Flash 0731
Claude 3 Haiku vs DeepSeek V4 Flash 0731
DeepSeek V4 Flash 0731 leads on 2 of 2 shared benchmarks. DeepSeek V4 Flash 0731 is 4.4× cheaper per token on a blended 3:1 basis. DeepSeek V4 Flash 0731 accepts the larger context window (1.0M).
Head to head
Blended price uses a 3:1 input:output mix
| Claude 3 Haiku | DeepSeek V4 Flash 0731 | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Input $/1M | $0.250 | $0.090 |
| Output $/1M | $1.25 | $0.180 |
| Blended $/1M | $0.500 | $0.113 |
| Context window | 200K | 1.0M |
| Max output | 4K | 384K |
| Released | Mar 2024 | Jul 2026 |
| Licence | Proprietary | Open weights |
| GPQA Diamond | 36.3% | 91.0% |
| OTIS Mock AIME | 1.8% | 94.4% |
Benchmark scores independently run and published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0. Only benchmarks with a score for BOTH models are compared.
Choose Claude 3 Haiku if…
- No clear advantage on the data we hold.
Choose DeepSeek V4 Flash 0731 if…
- Costs less per token — $0.113 vs $0.500 blended
- Higher GPQA Diamond (91.0% vs 36.3%)
- Higher OTIS Mock AIME (94.4% vs 1.8%)
- Larger context window (1.0M)
- Open weights — self-hostable
- Newer model (released Jul 2026)
FAQ
Which is better, Claude 3 Haiku or DeepSeek V4 Flash 0731?
Across the 2 benchmarks both models have independently-run scores for, Claude 3 Haiku leads on 0 and DeepSeek V4 Flash 0731 leads on 2. Benchmarks are one input — price and context window are on this page too.
Is Claude 3 Haiku cheaper than DeepSeek V4 Flash 0731?
DeepSeek V4 Flash 0731 is cheaper: $0.113 versus $0.500 per million tokens blended at 3:1 input:output. Input alone: $0.250 vs $0.090. Output alone: $1.25 vs $0.180.
What context windows do Claude 3 Haiku and DeepSeek V4 Flash 0731 have?
Claude 3 Haiku: 200,000 tokens. DeepSeek V4 Flash 0731: 1,048,576 tokens.
Who makes Claude 3 Haiku and DeepSeek V4 Flash 0731?
Claude 3 Haiku is made by Anthropic. DeepSeek V4 Flash 0731 is made by DeepSeek.