DeepSeek-V4.1-Flash
552B MoE with 8B-active prefill / 16B-active decode. CED encoder-decoder roughly halves prefill; CSA2 cross-layer KV sharing plus FP4 shrink global KV to ~1/4 and persistent KV to ~1/8. Price/spec fields are estimated from the V4-Flash tier pending the full model card.
01
Snapshot
Overall
76
rank #12 of 27
Coding
89
rank #5 of 27
Multimodal
78
rank #16 of 27
Input / Output
$0.14 / $0.28
USD per 1M tokens
Context
1M
usable recall 98
Speed
~85 tok/s
TTFT 0.3s
Aggregated from public sources and independently weighted; methodology on the Terms page. Scores within 3 points are treated as statistically tied.
02
Specs & pricing
Every tracked field. List API prices in USD per 1M tokens; verify current vendor pricing before purchase.
| Vendor | DeepSeek (CN) |
| Released / current | 2026.09 |
| Licensing | Open weights |
| Context window | 1M |
| Max output | 128K |
| Effective-context score | 98 |
| Input / output price | $0.14 / $0.28 per 1M tokens |
| Cache discount | No published cache discount |
| Free tier | DeepSeek App free; open weights to self-host |
| Speed / TTFT | ~85 tok/s / 0.3s |
| Function calling | 84 |
| Refusal rate | ~5% |
| English / Chinese | 84 / 86 |
| Modalities | text, image |
| Fine-tuning | Yes |
| Private deployment | Yes (enterprise) |
| SOC2 / no-train | no / yes |
03
Strengths & watch-outs
Strengths
- 1M context at ~890 bytes/token global KV
- Top agentic coding (DeepSWE 74.2, Terminal-Bench 2.1 90.6)
- Persistent KV cut to ~1/8 via bounded replay
- Open weights, extremely low serving cost
Watch-outs
- Gap on hardest scientific agents (Terminal-Bench 4.0)
- Multimodal below top closed models
- Younger, smaller ecosystem
04
Where it ranks among 27 models
Overall 76/100 (#12), coding 89 (#5), multimodal 78 (#16). See the full boards and side-by-side compare on the leaderboard.
05
Pricing in practice
Illustrative monthly bill for 100M input + 30M output tokens: $22 at list. Your mix and cache-hit ratio change this.
Illustrative model using public list rates before any enterprise agreement; cache discount: No published cache discount.
06
Frequently asked questions
How much does DeepSeek-V4.1-Flash cost?
List API pricing is $0.14 input and $0.28 output per 1M tokens. No published cache discount. DeepSeek App free; open weights to self-host
What is DeepSeek-V4.1-Flash's context window, and is it usable end to end?
It advertises a 1M window with 128K max output; its effective-context score is 98/100, which is the better predictor of whether long-document details are actually retained.
Is DeepSeek-V4.1-Flash good for coding?
Its coding aggregate is 89/100, rank #5 of 27. That puts it in the top tier for agentic, multi-file engineering.
Can I self-host DeepSeek-V4.1-Flash?
Yes — it ships open weights and supports private deployment, so you can run it on your own infrastructure for data control; budget for GPU and operations.
How strong is it in Chinese and on multimodal inputs?
Chinese score is 86 versus English 84; multimodal aggregate is 78/100 (rank #16), accepting text, image.