DeepSeek API Price per Million Tokens vs Frontier Western Labs

Table: DeepSeek API Price per Million Tokens vs Frontier Western Labs for DeepSeek. 3 rows across Model, Input, cache miss ($/1M tokens), Output ($/1M tokens), Notes.
ModelInput, cache miss ($/1M tokens)Output ($/1M tokens)Notes
DeepSeek V4.1-Flash (deepseek-flash)$0.30 peak / $0.15 off-peak$1.20 peak / $0.60 off-peakModel version V4.1-Flash-1017; 1M context
DeepSeek V4-Pro (deepseek-v4-pro)$1.32 peak / $0.66 off-peak$3.96 peak / $1.98 off-peakModel version V4-Pro-0813; 1M context
Anthropic Claude Opus 4.7 (reference)$5$25Anthropic list price, one rate at all hours

DeepSeek-V3 Disclosed Training Cost

Table: DeepSeek-V3 Disclosed Training Cost for DeepSeek. 3 rows across Figure, Value, Source / status.
FigureValueSource / status
Official V3 training run (GPU-hours)2.788M H800 GPU-hoursDeepSeek-V3 technical report (disclosed): 2.664M pre-training, 119K context extension, 5K post-training
Official V3 training cost (at $2/GPU-hr)$5.576MDeepSeek-disclosed at an assumed H800 rental price of $2 per GPU-hour; excludes prior research and ablation experiments
DeepSeek-R1 training cost (at $2/GPU-hr)$294K (147K H800 GPU-hours)DeepSeek R1 paper, Table 7: R1-Zero $202K, SFT data creation $10K, R1 $82K

DeepSeek Model-Family Release Timeline and Parameter Scaling

Table: DeepSeek Model-Family Release Timeline and Parameter Scaling for DeepSeek. 6 rows across Date, Model, Total / active params, context, Significance.
DateModelTotal / active params, contextSignificance
Nov 2023DeepSeek LLM (v1)7B and 67B parameters, trained on 2T tokensOpen-source Base and Chat variants
May 2024DeepSeek-V2236B total / 21B active, 128K contextIntroduced Multi-head Latent Attention (MLA) and DeepSeekMoE; released 6 May 2024
Dec 2024DeepSeek-V3671B total / 37B active, trained on 14.8T tokensTechnical report puts the official training run at 2.788M H800 GPU-hours, $5.576M at an assumed $2 per GPU-hour
Jan 2025DeepSeek-R1671B total / 37B activeOpen-weights reasoning model released 20 Jan 2025 under the MIT License; DeepSeek says its performance is on par with OpenAI-o1; 90.8 MMLU in the R1 paper
Dec 2025DeepSeek-V3.2Continued pre-training from V3.1-Terminus (128K context)Released 1 Dec 2025 with V3.2-Speciale; first DeepSeek model to integrate thinking directly into tool use
Apr 24 2026DeepSeek-V4 (Pro + Flash)V4-Pro 1.6T / 49B active; V4-Flash 284B / 13B active; 1M contextOpen-sourced preview; V4-Pro pre-trained on 33T tokens and V4-Flash on 32T; DeepSeek says V4-Pro rivals the world's top closed-source models
DeepSeek in AI Software cover

DeepSeek in AI Software

KPI data tables

DeepSeek operating metrics behind the KPI charts, newest period first (up to 24 periods per chart). Periods use each company's own fiscal labels where the source does; each table names its source.

Competitive-Programming Rating: DeepSeek V4-Pro vs Frontier Models (Codeforces) (Codeforces rating)
PeriodDeepSeek V4-Pro MaxOpenAI GPT-5.4 xHighGoogle Gemini-3.1-Pro High
Codeforces rating3,2063,1683,052

Source: DeepSeek-V4-Pro model card, DeepSeek-V4-Pro-Max vs Frontier Models table (huggingface.co/deepseek-ai/DeepSeek-V4-Pro); DeepSeek-V4 Technical Report (arxiv.org/abs/2606.19348); DeepSeek API Docs, DeepSeek-V4 Preview Release (24 Apr 2026).

DeepSeek V4-Pro vs Frontier Models on Coding Benchmarks (score)
PeriodDeepSeek V4-Pro MaxClaude Opus-4.6 MaxGoogle Gemini-3.1-Pro High
LiveCodeBench93.588.891.7
SWE-bench Verified (%)80.680.880.6

Source: DeepSeek-V4-Pro model card, DeepSeek-V4-Pro-Max vs Frontier Models table (huggingface.co/deepseek-ai/DeepSeek-V4-Pro); DeepSeek-V4 Technical Report (arxiv.org/abs/2606.19348); DeepSeek API Docs, DeepSeek-V4 Preview Release (24 Apr 2026).

Ask Sterling

Register for a premium account to gain access to Sterling AI.

Get Started

Things you can ask Sterling:

Summarize Tesla's latest earnings reportWhy did NVIDIA's margins expand?Compare Apple vs Microsoft's cash flowWhat's driving EV industry growth?
Menu
Favorites
DeepSeek - KPIs | Sterling