DeepSeek V4 Flash Vision Exp
DeepSeekReleased Aug 21, 2026Open weights · MITdeepseek-v4-flash-vision-exp
- Reasoning
- yes, with effort levels
- Knowledge cutoff
- March 2026
- Context
- 1M tokens
- Max output
- 393k tokens
- Input
- text
- API features
- tools, structured outputs, JSON mode, seed, logprobs
- DeepSeek's API
- yes, as deepseek-v4-flash-vision-exp
Read Oct 23, 06:00 UTC from the hosts' published offers and OpenRouter, spec sheet from DeepSeek · cutoff: aiknowledgecutoff.com
Speed by provider
Output speed of DeepSeek V4 Flash Vision Exp at each host, fastest first
- DeepInfra123 tok/s1.02 s to first token
- SiliconFlow94 tok/s1.28 s to first token
- GMICloud85 tok/s3.09 s to first token
Speed: OpenRouter, median of the last 30 min, read Oct 23, 06:00 UTC · a snapshot of real traffic before the daily read, not a daily average · bars from zero.
- Parameters
- 284B
- 13B active per token (MoE)
- Memory needed
- 170 GB
- in INT4 · 341 GB in INT8
- Smallest cloud setup
- 4× H100
- 320 GB · $8.60/h, about $6,278 a month
- Worth it above
- 7B tokens
- a month, against the cheapest API at $0.86 per M
| Quantization | Memory | Smallest cloud GPU | Quality |
|---|---|---|---|
| INT4 (Q4) | 170 GB | 4× H100 · $8.60/h | smallest footprint, minor quality loss |
| INT8 (Q8) | 341 GB | 8× H100 · $17.20/h | near-lossless |
| FP16 (full) | 682 GB | multi-node, over 640 GB | reference quality |
At home: in INT4 it fits a Mac Studio M5 Ultra (256 GB unified), about $9,499 once, or $0.3615/h over 3 years, electricity aside; it pays for itself after about 11B tokens against the cheapest API.
Memory ≈ parameters × bytes per parameter × 1.2 · cloud price: on-demand rate, running 24/7 · builds: GGUF, AWQ, MLX, GPTQ, Ollama · break-even calculator
| Provider | Input | Output | Cache ↑ | tok/s | First token | Uptime 24h | Checked |
|---|---|---|---|---|---|---|---|
| DeepInfrafp8 · 1M ctx · 256K out−51% | $0.22 | $0.65 | $0.0069 | 123 | 1.02 s | 100% | Oct 23 |
| Fireworks AI1M ctx | $0.22 | $0.66 | $0.007 | — | — | 98.9% | Sep 26 |
| DeepSeekprecision not stated · 1M ctx · 375K out | $0.22 | $0.66 | $0.007 | — | — | — | Sep 9 |
| GMICloudfp8 · 1.05M ctx | $0.44 | $1.32 | $0.014 | 85 | 3.09 s | 99.9% | Oct 23 |
| Novita AIprecision not stated · 1M ctx · 384K out | $0.44 | $1.32 | $0.028 | — | — | — | Oct 8 |
| SiliconFlowfp8 · 1M ctx · 384K out | $0.44 | $1.32 | $0.028 | 94 | 1.28 s | 100% | Oct 23 |
| AtlasCloudprecision not read: host terms · 1M ctx · 64K out | $0.44 | $1.32 | $0.028 | — | — | 79.7% | Sep 15 |
7 of 7 providers · prices per million tokens, read Oct 23, 06:00 UTC · speed: OpenRouter, median of the last 30 min, read Oct 23, 06:00 UTC · uptime: OpenRouter, last 24 h · −N%: a discount off the host's list price · under each host: weights precision, context and max output as served, the maker's value in brackets when the host's differs
| Provider | Weights | Context | Max output | Tools | Structured outputs | Reasoning |
|---|---|---|---|---|---|---|
| DeepInfra | fp8 | 1M | 256K | yes | yes | yes |
| Fireworks AI | not checked yet | 1M | not checked yet | — | — | — |
| DeepSeek | not stated by the host | 1M | 375K | — | — | — |
| Novita AI | not stated by the host | 1M | 384K | — | — | — |
| SiliconFlow | fp8 | 1M | 384K | yes | no | yes |
| AtlasCloud | not read: host terms | 1M | 64K | — | — | — |
| GMICloud | fp8 | 1.05M | not read: host terms | yes | no | yes |
Weights, context and max output: read at each host's own catalogue or docs where its terms allow, otherwise as the host declares to OpenRouter; hover a cell for its source and date · tools, structured outputs, reasoning: as declared to OpenRouter · not verified by us
DeepSeek V4 Flash Vision Exp spec sheet, field by field, with sources
DeepSeek V4 Flash Vision Exp spec sheetmaker-declared
What DeepSeek declares about the model itself — read from its own weights repository and documentation, not from any host. What each endpoint actually serves is in the pricing above: the two can differ, and both are true.
- Context window
- 1,048,576 tokens
- Parameters
- 304,646,824,126≈ 304.6B
- Active experts
- 6 of 256 active per token
- Weight precision
- fp8
- License
- MIT — open weights, permissivecurated · read 2026-09-07Scope: the licence on the published weights, and nothing else. It is not the maker's acceptable-use policy, which is a separate document, and it is not the contract you sign with whichever host you call — hosts set their own terms, which this index does not read.
- Knowledge cutoff
- not published by the maker≈ March 2026 third-party estimate — not the maker's figurecurated
- Modalities
- text
- Max output (official)
- to verify
- Release date
- to verify≈ August 21, 2026 third-party catalogue — not the maker's figure
- Maker lifecycle status
- Retired · superseded by DeepSeek V4.1 Flashmaker's docs · read 2026-09-10
Also declared in the weights
- Architecture class
- DeepseekV4ForCausalLM
- Layers
- 43
- Vocabulary
- 129,280 tokens
- Weights first published
- August 31, 2026
- Weights last updated
- September 1, 2026
deepseek-ai/DeepSeek-V4-Flash-Vision-Exp · config.json @ 6821d6a · read 2026-10-08 · 7 of 10 fields published by the maker
| Provider | Chatbot 1,000 in / 500 out | RAG / long context 8,000 in / 500 out | Batch summarize 4,000 in / 1,000 out |
|---|---|---|---|
| DeepInfra | $0.54 | $2.05 | $1.51 |
| Fireworks AI | $0.55 | $2.09 | $1.54 |
| DeepSeek | $0.55 | $2.09 | $1.54 |
| Novita AI | $1.10 | $4.18 | $3.08 |
| SiliconFlow | $1.10 | $4.18 | $3.08 |
| AtlasCloud | $1.10 | $4.18 | $3.08 |
| GMICloud | $1.10 | $4.18 | $3.08 |
What each host bills for 1,000 requests of a representative job
Your own workload
Other DeepSeek models · all its models, hosts and history ›
- DeepSeek V4.1 Flash29 providers · from $0.47
- DeepSeek V4 Flash 073125 providers · from $0.21
- DeepSeek V4 Pro 081323 providers · from $2.50
- DeepSeek V4.1 Flash 101722 providers · from $0.39
- DeepSeek V4 Pro 042315 providers · from $2.40
- DeepSeek V3.214 providers · from $0.52