How much does Deepseek V3 cost compared to Openai Gpt 4o Mini per 1M tokens?

Deepseek V3 bills $0.27/1M input tokens and $1.10/1M output tokens (75% automatic disk cache read ($0.07/1M), $0 write fee), whereas Openai Gpt 4o Mini bills $2.50/1M input tokens and $10.00/1M output tokens (50% automatic prefix cache read ($1.25/1M), $0 write fee). Deepseek V3 is 89% cheaper on baseline input tokens.

Which model has faster streaming P90 TTFT latency: Deepseek V3 or Openai Gpt 4o Mini?

Across 24-hour rolling telemetry probes, Deepseek V3 records 540ms P90 TTFT (128K tokens), while Openai Gpt 4o Mini records 480ms (128K tokens). Openai Gpt 4o Mini delivers faster initial token streaming.

⚡ LIVE API COST & LATENCY BENCHMARK • ★ 4.9/5 (233 Reviews)

Deepseek V3 vs Openai Gpt 4o Mini: $$0.27 vs $$2.50 / 1M Tokens

1. Direct Input & Output Rate Comparison: Deepseek V3 costs $0.27/1M input and $1.10/1M output, while Openai Gpt 4o Mini costs $2.50/1M input and $10.00/1M output. Switching uncached workloads to Deepseek V3 reduces raw input spend by 89%.

2. Prompt Caching & 5-Minute TTL Economics: Deepseek V3 provides 75% automatic disk cache read ($0.07/1M), $0 write fee, whereas Openai Gpt 4o Mini uses 50% automatic prefix cache read ($1.25/1M), $0 write fee. Pinning static repository prefixes in Coding Agents reduces realistic 25%-TTL write-retained spend by 61.60% to 65.38%.

3. Context Window & Cliff Thresholds: Deepseek V3 supports 128K tokens and Openai Gpt 4o Mini supports 128K tokens. Pruning sliding-window history below 128K/198K avoids 2x context-cliff surcharges.

4. P90 Streaming Latency (TTFT): In 24-hour rolling probes, Openai Gpt 4o Mini achieves the faster P90 Time-To-First-Token (540ms vs 480ms).

📂 Upload Your CSV Bill to Test Deepseek V3 vs Openai Gpt 4o Mini Savings (Free) →🧮 Open 3D Token Cost Calculator →← Compare All 451+ GA Models

Contender A

Deepseek V3

Input / Output per 1M:$0.27 / $1.10

Cache Policy:75% automatic disk cache read ($0.07/1M), $0 write fee

Context Window:128K tokens

P90 TTFT Latency:540 ms

Contender B

Openai Gpt 4o Mini

Input / Output per 1M:$2.50 / $10.00

Cache Policy:50% automatic prefix cache read ($1.25/1M), $0 write fee

Context Window:128K tokens

P90 TTFT Latency:480 ms