Claude Haiku 5.5 vs DeepSeek V4 Pro 0813
Anthropic vs Tencent Cloud (official direct) · prices and specs, line by line
Claude Haiku 5.5 (Anthropic) vs DeepSeek V4 Pro 0813 (Tencent Cloud (official direct)), side by side: input/output/cache-hit prices, time-based pricing, context window, modalities and a fixed-assumption monthly cost. Green marks the better row.
| Dimension | Claude Haiku 5.5 | DeepSeek V4 Pro 0813 |
|---|---|---|
| Vendor / channel | Anthropic | Tencent Cloud (official direct) |
| Context window | 1024K | 1000K |
| Modalities | Text, Image understanding | Text |
| Features | Low-cost, lightweight, Subtask agent | Flagship reasoning, Half price off-peak, Strong agent skills |
| Input (cache miss) | Prompt ≤100K $0.1 Prompt >100K $0.5 | Peak ¥9 / off-peak ¥4.5 |
| Output | Prompt ≤100K $0.5 Prompt >100K $2.5 | Peak ¥27 / off-peak ¥13.5 |
| Cache hit | Prompt ≤100K $0.01 Prompt >100K $0.05 | Peak ¥0.3 / off-peak ¥0.15 |
| Time-based pricing | Off-peak ≈ half of peak (peak=Weekdays 9:00-12:00, 14:00-18:00) | |
| Est. monthly cost* | $1.64 |
¥51.57 |
| Price source | Anthropic official pricing | Tencent Cloud TokenHub pricing |
| Verified | 2026-10-28 | 2026-09-18 |
* Fixed assumptions for the example monthly cost: 100 calls/day · 2,000 input / 800 output tokens per call · off-peak rates · 30% cache hit rate · 30-day month — the same formula and defaults as the cost calculator. USD models are converted to CNY at the reference rate for comparison (see summary). Costs within display precision are marked tied; neither side is favored.
Comparison summary (fixed usage basis)
- Estimated monthly cost: Claude Haiku 5.5 ¥11.00 vs DeepSeek V4 Pro 0813 ¥51.57 on a common CNY basis (list prices $1.64 / ¥51.57) — at this fixed usage, Claude Haiku 5.5 is cheaper by ¥40.57/month.
- Cost ratio: under this basis, DeepSeek V4 Pro 0813 costs about 4.7× Claude Haiku 5.5 (Claude Haiku 5.5 is ~79% cheaper). The ratio holds only at the current list prices, reference rate and fixed usage assumptions; it says nothing about capability, output quality or overall value.
- Input unit price (off-peak / single-price basis): list prices $0.1 vs ¥4.5; after conversion Claude Haiku 5.5 is lower (by ≈ ¥3.83/1M tokens).
- Output unit price (off-peak / single-price basis): list prices $0.5 vs ¥13.5; after conversion Claude Haiku 5.5 is lower (by ≈ ¥10.14/1M tokens).
- Cache-hit price (off-peak basis): Claude Haiku 5.5 $0.01; DeepSeek V4 Pro 0813 ¥0.15.
- Currencies: Claude Haiku 5.5 is billed in USD, DeepSeek V4 Pro 0813 in CNY; conversion uses the CFETS central parity rate (see above) and is indicative only.
- Context window: Claude Haiku 5.5 1024K vs DeepSeek V4 Pro 0813 1000K — Claude Haiku 5.5 is 1.0× DeepSeek V4 Pro 0813.
- Modalities: Claude Haiku 5.5=Text, Image understanding; DeepSeek V4 Pro 0813=Text (Claude Haiku 5.5 adds Image understanding).
- Price sources: Anthropic official pricing (verified 2026-10-28), Tencent Cloud TokenHub pricing (verified 2026-09-18); unit: official list price per 1M tokens (Claude Haiku 5.5 in USD, DeepSeek V4 Pro 0813 in CNY; cross-currency figures converted to CNY as noted above).
These are computations from published list prices under fixed call volume, token usage, cache assumptions and billing rules. They are not rankings of capability, output quality or overall value, and do not constitute purchase advice.