DeepSeek V4 Pro 0423
DeepSeekReleased Apr 24, 2026Open weights · MITdeepseek-v4-pro-0423
- Intelligence (Epoch)
- 149 · 44th of 111
- 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
Read Oct 23, 06:00 UTC from the hosts' published offers and OpenRouter, spec sheet from DeepSeek · cutoff: aiknowledgecutoff.com
The cheapest host, Relace, serves fp4 weights (maker: fp8). Cheapest at the maker's precision and context: StreamLake, $2.87 blended. Weights and context per host
Intelligence for the price
11 models score higher for the same price or less; it ranks 12 of the 57 models at its price or below.
Epoch Capabilities Index against the lowest input + output price per M, log scale · prices Oct 23. Score: Epoch AI, CC BY 4.0, 149 with an interval of 148 to 151.
Against its rivals
The best of its price range and the top score
- DeepSeek V4 Pro 0423this model$2.40149
- Claude Opus 5.5top score$24.00169
- Gemini 3.7 Flashsame price range$4.50158
- Gemini 3.8 Flashsame price range$4.50157
Same level for 13× less: GLM-5.3-Flash at StreamLake, $0.19 per M input + output · their Epoch intervals overlap (153 against 149).
Bars: Epoch Capabilities Index (Epoch AI, CC BY 4.0), drawn from 100, below the lowest score in the catalogue · price: lowest input + output per M.
Speed by provider
Output speed of DeepSeek V4 Pro 0423 at each host, fastest first (the 10 fastest of 13)
- Relace129 tok/s0.67 s to first token
- Azure OpenAI96 tok/s1.98 s to first token
- DeepInfra51 tok/s0.96 s to first token
- Alibaba46 tok/s1.82 s to first token
- Baidu43 tok/s0.91 s to first token
- Venice40 tok/s1.67 s to first token
- StreamLake39 tok/s1.45 s to first token
- Cloudflare39 tok/s1.66 s to first token
- AtlasCloud38 tok/s1.29 s to first token
- Reka34 tok/s0.69 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
- 1599B
- 49B active per token (MoE)
- Memory needed
- 959 GB
- in INT4 · 1919 GB in INT8
| Quantization | Memory | Smallest cloud GPU | Quality |
|---|---|---|---|
| INT4 (Q4) | 959 GB | multi-node, over 640 GB | smallest footprint, minor quality loss |
| INT8 (Q8) | 1919 GB | multi-node, over 640 GB | near-lossless |
| FP16 (full) | 3838 GB | multi-node, over 640 GB | reference quality |
At home: 959 GB in INT4 is beyond any desktop or Mac in our machines table, so it means a multi-GPU server.
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 |
|---|---|---|---|---|---|---|---|
| StreamLakefp8 · 1000K ctx · 375K out−45% | $0.96 | $1.91 | $0.0796 | 39 | 1.45 s | 100% | Oct 23 |
| GMICloudfp8 · 1M ctx · 384K out−45% | $0.96 | $1.91 | $0.0798 | 22 | 2.52 s | 99.9% | Oct 23 |
| DeepInfrafp8 · 1M ctx · 16K out | $1.30 | $2.60 | $0.10 | 51 | 0.96 s | 100% | Oct 23 |
| Parasailfp8 · 1M ctx · 384K out | $0.45 | $3.48 | $0.10 | — | — | 72.3% | Oct 22 |
| Alibabafp8 · 1M ctx · 384K out | $1.42 | $2.83 | $0.12 | 46 | 1.82 s | 98.8% | Oct 23 |
| AtlasCloudfp4 (fp8) · 1M ctx · 384K out | $1.68 | $3.38 | $0.13 | 38 | 1.29 s | 99.3% | Oct 23 |
| SiliconFlowfp8 · 1M ctx · 384K out | $1.50 | $3.13 | $0.14 | 34 | 2.12 s | 98.0% | Oct 23 |
| Novita AIfp8 · 1M ctx · 384K out | $1.60 | $3.20 | $0.14 | — | — | — | Oct 23 |
| Baidufp8 · 1M ctx · 384K out | $1.69 | $3.38 | $0.14 | 43 | 0.91 s | 100% | Oct 23 |
| Azure OpenAI1M ctx · 375K out | $1.91 | $3.83 | $0.16 | 96 | 1.98 s | 100% | Oct 23 |
| Cloudflare1M ctx · 384K out | $1.15 | $2.55 | $0.20 | 39 | 1.66 s | 99.9% | Oct 23 |
| Rekaprecision not read: host terms · 1M ctx · 384K out−30% | $1.05 | $2.10 | $0.21 | 34 | 0.69 s | 98.9% | Oct 23 |
| Relacefp4 (fp8) · 1M ctx · 384K out | $0.90 | $1.50 | $0.25 | 129 | 0.67 s | 100% | Oct 23 |
| Veniceprecision not read: host terms · 1M ctx · 32K out | $1.65 | $3.30 | $0.33 | 40 | 1.67 s | 98.7% | Oct 23 |
| DigitalOcean1M ctx · 375K out | $1.74 | $3.48 | $0.35 | 26 | 1.42 s | 100% | Oct 23 |
10 of 15 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 |
|---|---|---|---|---|---|---|
| Relace | fp4 | 1M | 384K | yes | no | yes |
| StreamLake | fp8 | 1000K | 375K | yes | yes | yes |
| GMICloud | fp8 | 1M | 384K | yes | no | yes |
| Reka | not read: host terms | 1M | 384K | yes | yes | yes |
| Cloudflare | not checked yet | 1M | 384K | yes | yes | yes |
| DeepInfra | fp8 | 1M | 16K | yes | yes | yes |
| Parasail | fp8 | 1M | 384K | yes | yes | yes |
| Alibaba | fp8 | 1M | 384K | yes | yes | yes |
| SiliconFlow | fp8 | 1M | 384K | yes | no | yes |
| Novita AI | fp8 | 1M | 384K | yes | no | yes |
| Venice | not read: host terms | 1M | 32K | yes | yes | yes |
| AtlasCloud | fp4 | 1M | 384K | yes | yes | yes |
| Baidu | fp8 | 1M | 384K | yes | yes | yes |
| DigitalOcean | not checked yet | 1M | 375K | yes | no | yes |
| Azure OpenAI | not checked yet | 1M | 375K | 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 Pro 0423 spec sheet, field by field, with sources
DeepSeek V4 Pro 0423 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
- 1,598,839,674,782≈ 1.60T
- Active experts
- 6 of 384 active per token
- Weight precision
- fp8
- License
- MIT — open weights, permissivecurated · read 2026-10-21Scope: 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
- April 24, 2026maker's docs · read 2026-10-21
- Maker lifecycle status
- Previewmaker's docs · read 2026-10-21
Also declared in the weights
- Architecture class
- DeepseekV4ForCausalLM
- Layers
- 61
- Vocabulary
- 129,280 tokens
- Weights first published
- April 22, 2026
- Weights last updated
- June 22, 2026
- Paper
- arXiv:2606.19348
deepseek-ai/DeepSeek-V4-Pro · config.json @ b5968e9 · read 2026-10-21 · 8 of 10 fields published by the maker
| Date | Provider | Input | Output | Change |
|---|---|---|---|---|
| Oct 23 | Baidu | $0.44 → $1.69 | $0.88 → $3.38 | +283% |
| Oct 23 | StreamLake | $0.44 → $0.96 | $0.89 → $1.91 | +116% |
Every published change of input or output price, newest first · change on input + output · full series: pricing-history.csv.
| Provider | Chatbot 1,000 in / 500 out | RAG / long context 8,000 in / 500 out | Batch summarize 4,000 in / 1,000 out |
|---|---|---|---|
| Relace | $1.65 | $7.95 | $5.10 |
| StreamLake | $1.91 | $8.60 | $5.73 |
| GMICloud | $1.91 | $8.61 | $5.74 |
| Reka | $2.10 | $9.45 | $6.30 |
| Cloudflare | $2.43 | $10.47 | $7.15 |
| DeepInfra | $2.60 | $11.70 | $7.80 |
| Parasail | $2.19 | $5.34 | $5.28 |
| Alibaba | $2.83 | $12.74 | $8.50 |
| SiliconFlow | $3.07 | $13.58 | $9.14 |
| Novita AI | $3.20 | $14.40 | $9.60 |
| Venice | $3.30 | $14.85 | $9.90 |
| AtlasCloud | $3.37 | $15.13 | $10.10 |
| Baidu | $3.38 | $15.21 | $10.14 |
| DigitalOcean | $3.48 | $15.66 | $10.44 |
| Azure OpenAI | $3.82 | $17.19 | $11.47 |
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 V3.214 providers · from $0.52
- DeepSeek V4 Flash 042312 providers · from $0.21