· CostMyTokens · Analysis · 3 min read
DeepSeek V4 Pro API Pricing vs GPT-6, Claude 5.5, Gemini 4
DeepSeek V4 Pro's official and gateway prices against the flagship closed models, with worked monthly costs at 10M and 100M tokens.
DeepSeek V4 Pro is one of the few open-weight models with an official API price from its maker, which makes it a clean comparison against the closed flagships. Every price below is per 1M tokens, carries its source and verification date, and the costs are computed from the same data as our calculator.
Disclosure: links to Novita AI are affiliate links — we may earn a commission at no extra cost to you. It never affects order: every table here is sorted by price only. Read our disclosure.
DeepSeek V4 Pro: three ways to pay
| Where | Input / 1M | Output / 1M | Cached input / 1M | Source |
|---|---|---|---|---|
| DeepSeek official API (off-peak, 50% off) | $0.66 | $1.98 | $0.022 | DeepSeek pricing, verified 2026-10-25 |
| DeepSeek official API (peak hours) | $1.32 | $3.96 | $0.044 | DeepSeek pricing, verified 2026-10-25 |
| Novita AI gateway (affiliate) | $1.32 | $3.96 | $0.044 | Novita pricing, verified 2026-10-25 |
Peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday–Friday (excluding Chinese public holidays); all other hours are off-peak at half price. At peak hours, Novita’s rate for the same model version (DeepSeek V4 Pro 0813) matches DeepSeek’s official rate exactly; off-peak, DeepSeek direct is half price.
Against the closed flagships
Workload: 10M tokens a month, 8M input / 2M output, no caching, standard tier. Sorted by monthly cost.
| Model | Price type | Input / 1M | Output / 1M | 10M tokens / month | 100M tokens / month | Source |
|---|---|---|---|---|---|---|
| DeepSeek V4 Pro — DeepSeek official API (off-peak, 50% off) | Official | $0.66 | $1.98 | $9.24 | $92.40 | DeepSeek pricing, verified 2026-10-25 |
| DeepSeek V4 Pro — DeepSeek official API (peak hours) | Official | $1.32 | $3.96 | $18.48 | $184.80 | DeepSeek pricing, verified 2026-10-25 |
| DeepSeek V4 Pro — Novita AI gateway (affiliate) | Gateway | $1.32 | $3.96 | $18.48 | $184.80 | Novita pricing, verified 2026-10-25 |
| Claude Sonnet 5.5 | Official | $2.0 | $10.0 | $36.00 | $360.00 | official list price (source), verified 2026-10-25 |
| GPT-6.1 Sol | Official | $2.0 | $10.0 | $36.00 | $360.00 | official list price (source), verified 2026-10-24 |
| Gemini 4 Argon | Official | $2.0 | $10.0 | $36.00 | $360.00 | official list price (source), verified 2026-10-24 |
| Gemini 3.1 Pro | Official | $2.0 | $12.0 | $40.00 | $400.00 | official list price (source), verified 2026-10-25 |
| Claude Opus 5.5 | Official | $4.0 | $20.0 | $72.00 | $720.00 | official list price (source), verified 2026-10-25 |
| GPT-6 Astra | Official | $10.0 | $50.0 | $180.00 | $1,800.00 | official list price (source), verified 2026-10-24 |
At this mix, DeepSeek V4 Pro at peak official rates costs $18.48 a month, versus $36.00 for GPT-6.1 Sol and $72.00 for Claude Opus 5.5 — that is 1.9x and 3.9x the DeepSeek bill. Off-peak, the DeepSeek figure halves to $9.24.
Caching narrows the gap — a little
DeepSeek bills cache hits at $0.044 per 1M at peak. For a workload of 8M input tokens of which 6M are cache hits, plus 2M output: DeepSeek V4 Pro (peak, official) costs $10.82; GPT-6.1 Sol (cached input $0.1) costs $24.60.
Caveats
- Price per token is not price per task: if a model needs more retries or longer outputs to finish a job, its effective cost rises. Benchmark on your own prompts.
- Gemini 4 Argon’s rate is introductory: Introductory pricing runs until Nov 30, 2026. From Dec 1, 2026, rates double to $4 / $20 per 1M.
- GPT-6.1 Sol: Prompts above ~272K input tokens enter a long-context pricing tier (about 2x input, 1.5x output). This calculator uses standard-tier rates.
Related
- DeepSeek V4 Pro cost calculator
- DeepSeek API pricing (all official models)
- Novita AI gateway prices
- OpenAI vs Anthropic vs Google pricing
- Estimate your own bill in the calculator
Disclosure: links to Novita AI are affiliate links — we may earn a commission at no extra cost to you. It never affects order: every table here is sorted by price only. Read our disclosure.