An independent field record

DeepSeek’s models, pricing, and claims — checked against the source.

An unofficial guide and directory to DeepSeek: the models (V4.1-Flash and V4-Pro), API prices, the DeepSeek Harness CLI (@deepseek-ai/dsh), news, and the open-source libraries. Independent resource, not affiliated with DeepSeek.

15 benchmark figures, each linked to its primary source.

Last updated 16 Nov 2026

DeepSeek-V4.1-Flash

September 2026
sourced figures
Context
1M tokens
1,048,576 Tokens (1M Native)
Throughput
Not published
No single tok/s figure on the model card.
Input / output
$0.30 / $1.20
USD per 1M tokens
552B MoE backbone8B prefill / 16B decodeEngram (196B)890-byte KV cache1M contextFP4 E2M1 KVAPI: deepseek-flashVision supported

View DeepSeek-V4.1-Flash in the models directory

Latest updates

See all news