Provider pricing hub
DeepSeek API Pricing (July 2026)
The DeepSeek API costs $0.435/$0.87 per million input/output tokens for DeepSeek V4 Pro and $0.14/$0.28 for V4 Flash. Cache-hit input falls to $0.003625 and $0.0028 respectively. Both V4 tiers ship a 1M-token context window.
DeepSeek is the price floor of the frontier-adjacent API market. The V4 generation kept the strategy that made V3 famous: one flagship (V4 Pro, $0.435/$0.87) priced below Western mid-tiers, one fast model (V4 Flash, $0.14/$0.28) priced below almost everything, and automatic context caching that makes repeated input cheaper still. We checked the official price table on July 14 and found the stored V4 Pro rate and cache fields had drifted. The table below now reads those exact values from the pricing registry.
Cheap only matters if the quality clears your bar: see the Chinese model rankings (where the V4 Pro reasoning rows lead), the price-vs-performance view, and the Chinese LLM deep dive.
Operator receipt
I checked DeepSeek's live pricing, cache, and concurrency documentation on July 15, 2026. At the calculator's default 10M input and 2M output tokens with no cache hits, V4 Pro costs $6.09 and V4 Flash costs $1.96. Those totals reproduce the registry rates instead of relying on a copied marketing example.
Honest limit. The estimate excludes taxes, retries, reasoning-token expansion, and any fallback provider. Cache savings use the hit share you enter; DeepSeek describes caching as best effort and does not guarantee a hit rate.
Reviewed by Glevd on July 15, 2026.
Every DeepSeek API price per 1M tokens
| Model | Input $/M | Cached input $/M | Output $/M | Context |
|---|---|---|---|---|
| DeepSeek R1 | $0.55 | $0.14 | $2.19 | 128K |
| DeepSeek V3.2 (Thinking) | $0.55 | $0.14 | $2.19 | 128K |
| DeepSeek V4 Pro | $0.435 | $0.004 | $0.87 | 1M |
| DeepSeek V3.2 | $0.28 | $0.028 | $0.42 | 128K |
| DeepSeek V3 | $0.27 | $0.07 | $1.1 | 128K |
| DeepSeek V4 Flash | $0.14 | $0.003 | $0.28 | 1M |
Cache-hit input rates are DeepSeek's official published prices (cache misses bill at the standard input rate). DeepSeek does not currently offer a Batch API discount tier; reasoning-effort variants (High/Max) inherit their family's token pricing and are collapsed into one row.
Estimate your monthly DeepSeek API bill
| Model | Est. monthly cost |
|---|---|
| DeepSeek R1 | $9.88 |
| DeepSeek V3.2 (Thinking) | $9.88 |
| DeepSeek V4 Pro | $6.09 |
| DeepSeek V3.2 | $3.64 |
| DeepSeek V3 | $4.9 |
| DeepSeek V4 Flash | $1.96 |
Estimates use DeepSeek's standard per-token rates from the table above. For task-level presets and cross-provider comparison, use the full AI cost calculator.
DeepSeek API quickstart: one authenticated request
DeepSeek exposes an OpenAI-compatible base URL at https://api.deepseek.com. Create a key in the DeepSeek platform, keep it on the server, and send it as a Bearer token. This request uses the current Flash model name rather than the retiring deepseek-chat alias:
curl https://api.deepseek.com/chat/completions \
-H 'Content-Type: application/json' \
-H "Authorization: Bearer $DEEPSEEK_API_KEY" \
-d '{"model":"deepseek-v4-flash","messages":[{"role":"user","content":"Reply with OK"}]}'A successful response includes token usage. Do not place the API key in browser code, logs, or a public repository. The cost calculator can then turn measured input, output, and cache-hit tokens into a monthly estimate.
Cache-hit pricing: DeepSeek’s biggest discount is automatic
DeepSeek prices input tokens in two lanes: cache miss (the standard rate) and cache hit, billed when the service can reuse a prompt prefix. A V4 Flash cache hit costs $0.0028 per million tokens; V4 Pro costs $0.003625. Those are explicit provider rates, not a percentage inferred from the standard input price.
Agents, RAG pipelines, and chat applications can benefit when they resend the same system prompt or document context. But the table does not promise a hit rate: a workload with mostly new prefixes pays the cache-miss rate. Enter your measured cached share in the estimator above. DeepSeek lists no separate Batch API tier.
The response exposes prompt_cache_hit_tokens and prompt_cache_miss_tokens in its usage object. Log those two fields and calculate the real hit share before budgeting. DeepSeek says unused cache entries are usually cleared after hours to days, so a one-time warm-up does not establish a permanent discount.
Concurrency and 429 handling
DeepSeek currently lists account-level concurrency limits of 500 for V4 Pro and 2,500 for V4 Flash. A request occupies one slot until its response completes. Requests beyond the account limit receive HTTP 429, regardless of how many API keys the account uses.
Production clients should cap concurrency, retry 429 responses with jittered backoff, and keep a fallback path for availability-sensitive work. DeepSeek also documents empty-line or SSE keep-alive traffic while a request waits; custom HTTP parsers must ignore those keep-alives without treating them as the JSON result.
V4 Pro vs V4 Flash: which DeepSeek tier to use
V4 Pro ($0.435/$0.87) is the flagship, with a 1M-token context window and separate reasoning-effort rows in the current Chinese model rankings. Use it for coding, reasoning, and agentic work where quality clears your workload's tests. V4 Flash ($0.14/$0.28) is the volume tier — classification, extraction, summarization, and high-throughput pipelines where per-request cost dominates.
Older V3.2 and R1rows remain in the registry for historical comparison. DeepSeek's current page says the legacy deepseek-chat and deepseek-reasoner aliases map to V4 Flash and are scheduled for deprecation on July 24, 2026. The V4 family is open weight, so you can also self-host instead of paying per token.
DeepSeek vs OpenAI and Claude pricing
The gap is not subtle. V4 Pro at $0.435/$0.87 costs about 17x less than GPT-5.6 Terra ($2.50/$15) on output and more than 28x less than GPT-5.6 Sol or Claude Opus 4.8 ($30 and $25 output respectively). V4 Flash at $0.14/$0.28 undercuts even OpenAI's nano tier on output. The honest caveat: the Western flagships still score higher on BenchLM's overall leaderboard, so the decision is quality-per-dollar, not price alone.
Compare the full tables: OpenAI API pricing and Claude API pricing, or the head-to-head Opus 4.8 vs V4 Pro comparison on the compare hub.
DeepSeek API pricing FAQ
How much does the DeepSeek API cost?
DeepSeek V4 Pro costs $0.435 input / $0.87 output per million tokens and V4 Flash costs $0.14/$0.28. Cache-hit input is $0.003625 and $0.0028 respectively. The current provider table maps legacy deepseek-chat and deepseek-reasoner aliases to V4 Flash and lists no separate batch tier.
Is the DeepSeek API free?
No — the hosted API is pay-per-token, though at prices low enough that light usage costs cents per day. The models themselves are open weight, so the genuinely free path is self-hosting a distilled or quantized variant on your own hardware; see the open-source model rankings.
What is DeepSeek cache-hit pricing?
When your request reuses a recently seen prompt prefix, DeepSeek bills that portion at the cache-hit rate instead of the standard input rate — $0.0028/M on V4 Flash and $0.003625/M on V4 Pro. The service applies caching automatically, but the hit share depends on repeated prompt prefixes. Use measured usage rather than assuming every agent or RAG workload reaches the same discount.
How does DeepSeek API pricing compare to OpenAI?
DeepSeek V4 Pro's output tokens cost about 34x less than GPT-5.6 Sol's ($0.87 vs $30 per million), and V4 Flash undercuts GPT-5.6 Luna by ~21x on output ($0.28 vs $6). OpenAI's flagships still lead on BenchLM's overall scores, so the right comparison is quality-per-dollar for your specific workload — see the OpenAI pricing hub.