DeepSeek V4.1 Flash (Together AI) API Pricing & Cost Rules
API Model ID: deepseek-ai/DeepSeek-V4.1-Flash · Context Window: 1,000,000 tokens · Max Output: 131,072 tokens
Context Tiers & Token Rate Card (USD per 1M Tokens)
| Tier | Std Input | Std Output | Cache Read | Write (Short) | Write (Long) |
|---|---|---|---|---|---|
| Standard (<= 1M) | $0.3 | $1.2 | $0.006 | $0.3 | N/A |
Prompt Caching Rules
Supported: Yes (automatic)
Minimum Cacheable Prefix: None / Automatic
Hourly Cache Storage: $0
Automatic fleet-shared KV cache (~1h best-effort TTL, subject to eviction under load). Read = $0.006/1M (98% off), zero write surcharge.
Batch API & Hosting Notes
Batch Supported: Yes (0% discount · 24h best-effort)
Caching in Batch: Allowed at standard cache rates (no batch discount on cache)
CRITICAL EXCEPTION: Together Batch API gives 50% off ONLY on openai/whisper-large-v3; all serverless chat models run at standard rates (0% discount, separate rate-limit pool only).
Served in FP8 quantization. US-only inference endpoint adds 1.25x (+25%) to input, cached input, and output.