Claude 3 Haiku is 1.4× cheaper per million tokens (blended). DeepSeek V4 Pro 0423 has the larger context window (1,024,000 vs 200,000 tokens, 5.1×). DeepSeek V4 Pro 0423 is open-weights; the other is not.
Claude 3 Haiku vs DeepSeek V4 Pro 0423
Live catalog fields. Best-in-row is highlighted.
Claude 3 Haiku
Anthropic · anthropic__claude-3-haiku
DeepSeek V4 Pro 0423open
DeepSeek · deepseek__deepseek-v4-pro
Identity
| Field | anthropic__claude-3-haiku | deepseek__deepseek-v4-pro |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Slug | anthropic__claude-3-haiku | deepseek__deepseek-v4-pro |
| Status | live | live |
| Open weights | No | Yes |
| License | — | — |
| HuggingFace | — | deepseek-ai/DeepSeek-V4-Pro |
Quality
| Field | anthropic__claude-3-haiku | deepseek__deepseek-v4-pro |
|---|---|---|
| Quality index (0–100) | — | 88.2 |
Cost
| Field | anthropic__claude-3-haiku | deepseek__deepseek-v4-pro |
|---|---|---|
| Input $ / 1M tokens | $0.25 | $0.72 |
| Cached input $ / 1M | $0.03 | $0.06 |
| Output $ / 1M tokens | $1.25 | $1.44 |
Context
| Field | anthropic__claude-3-haiku | deepseek__deepseek-v4-pro |
|---|---|---|
| Max context tokens | 200,000 | 1,024,000 |
| Max output tokens | 4,096 | 384,000 |
Modalities
| Field | anthropic__claude-3-haiku | deepseek__deepseek-v4-pro |
|---|---|---|
| Input | text, image | text |
| Output | text | text |