official model vendor provider · OpenAI-compatible with 7 normalized prices.
External public summaries normalized by LLMTicker; not LLMTicker verified user reviews. LLMTicker recalculates reputation from public items instead of copying third-party ratings.
DeepSeek reputation brief is based on 13 samples across 4 source families, with latest item at 2026-05-16T00:00:00.000Z; it helps scanning and does not replace item-level evidence.
Gateway user thread calls DeepSeek secondary-provider latency unacceptable for production traffic when routing balance runs out.
Gateway user thread calls DeepSeek secondary-provider latency unacceptable for production traffic when routing balance runs out.
Reliability note aggregates community concerns about DeepSeek request failures and slow response times after demand spikes.
Reliability note aggregates community concerns about DeepSeek request failures and slow response times after demand spikes.
Monitoring page tracks DeepSeek API timeout, slow, auth, and error reports, separating official API from third-party DeepSeek hosting.
Monitoring page tracks DeepSeek API timeout, slow, auth, and error reports, separating official API from third-party DeepSeek hosting.
Recent community thread discusses DeepSeek API being down during an update window, keeping availability risk visible for direct API users.
Recent community thread discusses DeepSeek API being down during an update window, keeping availability risk visible for direct API users.
Hands-on review frames DeepSeek API as a clear price-performance story for frontier-adjacent workloads, while production fit still depends on stability needs.
Hands-on review frames DeepSeek API as a clear price-performance story for frontier-adjacent workloads, while production fit still depends on stability needs.
GitHub issue requests clearer embedding-aware API support, showing developers still depend on issue threads for capability clarity.
GitHub issue requests clearer embedding-aware API support, showing developers still depend on issue threads for capability clarity.
HN discussion says DeepSeek-hosted models in China can have high latency, distinguishing model quality from hosted API experience.
HN discussion says DeepSeek-hosted models in China can have high latency, distinguishing model quality from hosted API experience.
User thread says DeepSeek API responses ranged from 30 to 80 seconds, raising overload and infrastructure concerns.
User thread says DeepSeek API responses ranged from 30 to 80 seconds, raising overload and infrastructure concerns.
| Model | Input miss | Input hit | Output | Verified | Source |
|---|---|---|---|---|---|
| $0.55/ 1M missFresh | Not published | $2.19/ 1M outFresh | 4 h | LiteLLM model cost map4 h | |
| $0.43/ 1M missFresh | $0.00/ 1M hitFresh | $0.87/ 1M outFresh | 4 h | DeepSeek pricing4 h | |
| $0.28/ 1M missFresh | $0.03/ 1M hitFresh | $0.42/ 1M outFresh | 4 h | LiteLLM model cost map4 h | |
| $0.28/ 1M missFresh | $0.03/ 1M hitFresh | $0.42/ 1M outFresh | 4 h | LiteLLM model cost map4 h | |
| $0.28/ 1M missFresh | Not published | $0.40/ 1M outFresh | 4 h | LiteLLM model cost map4 h | |
| $0.14/ 1M missFresh | Not published | $0.28/ 1M outFresh | 4 h | LiteLLM model cost map4 h | |
| $0.14/ 1M missFresh | $0.00/ 1M hitFresh | $0.28/ 1M outFresh | 4 h | DeepSeek pricing4 h |