// AI Platform Drift
DeepSeek V4 Flash pricing cut 33%, output limit raised to 384K tokens
OpenRouter model catalogue · published 2026-10-02 21:57 UTC
WHAT CHANGEDOpenRouter's DeepSeek V4 Flash 0423 model dropped from $0.0419/M prompt tokens and $0.0837/M completion tokens to $0.0280/M and $0.0560/M respectively, effective 2 October 2026. Maximum completion tokens increased from 131,072 to 384,000.
WHY IT MATTERSExisting workloads cost one-third less to run. Applications previously constrained by the 131K output limit can now generate responses nearly three times longer without chunking.
WHAT TO DOReforecast spend for any service using deepseek/deepseek-v4-flash and notify finance. If you truncated outputs or split requests to stay under 131K tokens, test whether removing that logic improves results.
Model-reported confidence 95% (not measured accuracy)
Primary source ↗
written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
id: deepseek/deepseek-v4-flash name: DeepSeek: DeepSeek V4 Flash 0423 -pricing.prompt: 0.00000004186 -pricing.completion: 0.00000008372 +pricing.prompt: 0.000000028 +pricing.completion: 0.000000056 context_length: 1048576 -top_provider.max_completion_tokens: 131072 +top_provider.max_completion_tokens: 384000
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
| triage verdict | MATERIAL at 0.98, needed 0.5 |
|---|---|
| triage reasoning | Pricing and output limits are direct cost and capability changes affecting deployment decisions. |
| signals | pricing.prompt decreased ~33% · pricing.completion decreased ~33% · max_completion_tokens increased 193% (131072 → 384000) |
| suppression | — |
| triage model | anthropic/claude-haiku-4.5 ($0.00000) |
| writer model | anthropic/claude-sonnet-4.5 ($0.00000) |