DeepSeek · Mid-tier · 164K context window
$0.003550 per call at 5,000 input / 2,000 output tokens.
Per call
$0.003550
10K calls/mo
$35.5
Cache savings
74% off input
Batch savings
0% off all
Adjust token counts, volume, and discount toggles to model your code review spend. Prices verified daily against provider pricing pages.
Call DeepSeek V3 with the OpenAI Python SDK pointed at api.deepseek.com:
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.deepseek.com/v1",
api_key=os.environ["DEEPSEEK_API_KEY"],
)
response = client.chat.completions.create(
model="deepseek-v3",
max_tokens=2000,
messages=[{"role": "user", "content": "Your prompt here"}]
)
# Cost: ~$0.003550 per call at standard rates
# Input: $0.27/MTok
# Output: $1.10/MTokPer-call cost for all models scoring 50+ on code review benchmarks. Hover for details.
Highest-quality affordable option: Claude Sonnet 5 at $0.0300/call (8.5x the cost, 96/100 quality). DeepSeek V3 is already the cheapest choice for code review.
Review is input-heavy in a way generation is not: the diff or file goes in whole, and structured feedback comes back short. That flips the usual optimization: input rate and cache discounts matter more than output rate. Reviewing the same repository repeatedly means the surrounding context is identical call over call, which is exactly the shape prompt caching is designed for.
For DeepSeek V3 specifically, that means output tokens cost 4.1x more than input, at $1.10 vs $0.27 per MTok. At this use case’s 5,000/2,000 token split, output accounts for 62% of the $0.003550 per-call cost.
| Direction | $/MTok | Tokens per call |
|---|---|---|
| Input | $0.27 | 5,000 |
| Output | $1.10 | 2,000 |
Caching drops input cost from $0.27 to $0.07/MTok. For repeated system prompts, that saves $0.001000 per call.
Submit requests asynchronously and receive 0% off all token costs. This works best for offline pipelines where latency requirements are flexible.
| Monthly volume | Cost per call | Standard monthly | Cached monthly | You save with caching |
|---|---|---|---|---|
| 10,000 calls | $0.003550 | $35.5 | $25.5 | $10 |
| 50,000 calls | $0.003550 | $177.5 | $127.48 | $50.02 |
| 100,000 calls | $0.003550 | $355 | $254.97 | $100.04 |
| 500,000 calls | $0.003550 | $1,775 | $1,274.83 | $500.17 |
| 1,000,000 calls | $0.003550 | $3,550 | $2,549.65 | $1,000.35 |
Benchmark data for code review is pending for DeepSeek V3. Test against your own evaluation suite before moving to production.
Benchmark data is not yet available for DeepSeek V3. We recommend testing against your own evaluation criteria.
6 models score 80+ on code review benchmarks, sorted by input rate. Cheapest: Claude Sonnet 5 (745% more, 96/100 quality).
| Model | Provider | Input $/MTok | Output $/MTok | Quality | Per call | vs DeepSeek |
|---|---|---|---|---|---|---|
| Claude Sonnet 5 | Claude | $2.00 | $10.00 | 96/100 | $0.0300 | 745% more |
| Claude Sonnet 4.6 | Claude | $3.00 | $15.00 | 95/100 | $0.0450 | 1168% more |
| GPT 5.6 Sol | GPT | $4.00 | $20.00 | 92/100 | $0.0600 | 1590% more |
| Claude Opus 4.8 | Claude | $5.00 | $25.00 | 94/100 | $0.0750 | 2013% more |
| GPT 5.5 | GPT | $5.00 | $30.00 | 90/100 | $0.0850 | 2294% more |
| Claude Fable 5 | Claude | $10.00 | $50.00 | 95/100 | $0.1500 | 4125% more |
Provider
DeepSeek
Tier
Mid-tier
Context
164K tokens
Released
Dec 2024
Prices are off-peak rates from api-docs.deepseek.com (read 2026-10-11). Peak-hour prices are 2x the listed input and output rates — Tokeven does not model time-of-day tiers; off-peak is used as the base. Cache hit: $0.07/MTok input. No batch discount published. Not yet benchmarked, so never recommended.
| Model | Tier | Input $/MTok | Output $/MTok | Context |
|---|---|---|---|---|
| DeepSeek R1 | Mid-tier | $0.55 | $2.19 | 128K |
| DeepSeek Flash | Fast / Budget | $0.15 | $0.60 | 1M |
| DeepSeek V4 Pro | Flagship | $0.66 | $1.98 | 1M |
You can compare DeepSeek V3 head-to-head against any other model on our model comparison pages, or track how pricing has changed over time on the pricing history tracker.
See how DeepSeek V3 costs change across different workload types.
Code Generation
$0.004940/call
Unit Test Generation
$0.006310/call
SQL Generation
$0.001505/call
API Generation
$0.007410/call
Code Refactoring
$0.005480/call
PR Review
$0.005460/call
Security Review
$0.004370/call
Bug Detection
$0.003550/call
Document Summarization
$0.003800/call
PDF Summarization
$0.005700/call
Meeting Notes
$0.004360/call
Email Summarization
$0.001360/call
Chatbot
$0.001640/call
RAG Pipeline
$0.003270/call
Customer Support Bot
$0.001910/call
FAQ Bot
$0.001285/call
Data Extraction
$0.002180/call
JSON Extraction
$0.001690/call
Invoice Parsing
$0.001200/call
Resume Parsing
$0.001555/call
Chain of Thought
$0.006310/call
Legal Analysis
$0.007100/call
Math Reasoning
$0.003705/call
Financial Analysis
$0.006560/call
Content Moderation
$0.000245/call
Sentiment Analysis
$0.000245/call
Categorization
$0.000381/call
Intent Detection
$0.000191/call
Copywriting
$0.002335/call
Blog Writing
$0.009070/call
Marketing Copy
$0.003516/call
Product Descriptions
$0.001785/call
A typical code review call using DeepSeek V3 costs about $0.003550 per call. This estimate uses 5,000 input tokens and 2,000 output tokens at standard rates. Standard rates are $0.27 per million input tokens and $1.10 per million output tokens.
Our benchmark data does not yet include a code review ranking for DeepSeek V3. We recommend testing it against your own evaluation criteria before production usage.
The most affordable model for code review above 80/100 quality is Claude Sonnet 5. It comes from Claude at $2.00 per million input tokens. Output costs $10.00 per million tokens with a quality score of 96/100. Claude Sonnet 5 costs more per call than DeepSeek V3 but scores higher on quality benchmarks for this task type.
Three strategies lower DeepSeek V3 costs for this task type effectively. Prompt caching gives a 74% discount on input tokens when reusing system prompts. The batch API gives 0% off all tokens for asynchronous processing workloads. Prompt optimization reduces token counts through more concise instructions and context windows. Combining caching with batching brings your cost from $0.003550 to about $0.000919 per call.
DeepSeek has 3 other models alongside DeepSeek V3 today: DeepSeek R1 at $0.55/$2.19 per million tokens (mid-tier), DeepSeek Flash at $0.15/$0.60 per million tokens (fast / budget), DeepSeek V4 Pro at $0.66/$1.98 per million tokens (flagship). The right choice depends on whether your code review workload prioritizes quality, cost, or speed.
DeepSeek V3 supports a context window of 164K tokens per API call. This sets the maximum combined length of your input prompt and generated response. For code review, typical calls use 5,000 input and 2,000 output tokens. That is well within this limit, leaving room for longer prompts or multi-turn conversations.
This calculator estimates costs based on published rates only. Connect your API usage to see where your money goes and which models drive your spend.