MODEL PROOF
DeepSeek V4 Pro 0423
BETTER VALUE OPTION
A stored alternative has an equal-or-higher measured score and an equal-or-lower price.
Serving precision differs between offers.
GLM 5.3 · Δ score 30.7 points · 9% lower measured price · $1.00 / 1M tokens
MiMo-V2.6-Pro is a further stored option with these losses: 393K → 131K max output
- Independent LMArena score
- 1,444.7 ± 6.63 · 9,544 votes
- Context window
- 1.05M
- Maximum output
- 393.22K
- Input modalities
- text
- Output modalities
- text
- Published input price
- $0.900 / 1M tokens
- Published output price
- $1.70 / 1M tokens
- Pricing kind
- fixed
Inspect complete billing conditions and endpoint terms below.
Price from SiliconFlow: the cheapest offer that is serving, at standard delivery, with at least 90% of the reference context and at a declared precision that is not 4-bit (fp8), at its standard rate. The maker’s own precision is not known.
Cheaper right now: $0.826/M at Baidu — promotion 61% It does not pass the like-for-like test, so it never sets a rank.
Every price dimension · USD per million tokens
| Input | $0.900/M |
|---|---|
| Output | $1.70/M |
| Cached input | $0.080/M |
| Cache write | Unknown/M |
| Cache write 1h | Unknown/M |
| Reasoning | Unknown/M |
Who sells it
The same weights, different shops. Cheapest is not like-for-like when serving precision differs.
Headline cheapest is a lower precision. Like-for-like at the best declared precision is $0.900/M from SiliconFlow.
13 provider offers · rates, limits and conditions
Rates are USD per million tokens. Endpoint terms can differ even when the quoted price and precision match. A listed parameter is a provider declaration, not a task-success test.
| Seller | Input | Output | Precision | Uptime (1d) |
|---|---|---|---|---|
| Baidubaidu/fp8 Endpoint terms
| $0.661 | $1.32 | fp8 | 99.93% |
| StreamLakestreamlake/fp8 Endpoint terms
| $0.662 | $1.32 | fp8 | 100.00% |
| SiliconFlowsiliconflow/fp8 Endpoint terms
| $0.900 | $1.70 | fp8 | 99.51% |
| Relacerelace/fp4 Endpoint terms
| $0.900 | $1.50 | fp4 | 100.00% |
| DeepInfradeepinfra/fp8/flex Endpoint terms
| $1.04 | $2.08 | fp8 | 100.00% |
| DigitalOceandigitalocean Endpoint terms
| $1.04 | $2.09 | undeclared | 100.00% |
| Novitanovita/fp8 Endpoint terms
| $1.28 | $2.56 | fp8 | 100.00% |
| DeepInfradeepinfra/fp8 Endpoint terms
| $1.30 | $2.60 | fp8 | 99.86% |
| Alibabaalibaba/fp8 Endpoint terms
| $1.42 | $2.83 | fp8 | 99.89% |
| Alibabaalibaba/fp8/us-east-1 Endpoint terms
| $1.56 | $3.12 | fp8 | 100.00% |
| Venicevenice Endpoint terms
| $1.65 | $3.30 | undeclared | 99.54% |
| Azureazure/us Endpoint terms
| $1.91 | $3.83 | undeclared | 100.00% |
| DeepInfradeepinfra/fp8/priority Endpoint terms
| $1.95 | $3.90 | fp8 | 100.00% |
Independent scores
| LMArena Elo | 1444.7high |
|---|---|
| LMArena row | deepseek-v4-pro-high-20260813 |
LMArena Elo under CC BY 4.0. Absence of another board is not a score of zero.
Best rankings by task
- Godot games #32 1059
- 3D #41 1220
- Web apps #46 1000
- Full-stack apps #48 948
- Game development #49 1204
- ASCII art #50 1134
- Code categories #55 1223
- SVG #60 1123
Per-task Elo and rank from Design Arena, via OpenRouter’s model feed. The rank is the feed’s, and it counts models OpenRouter does not list.
Capability
| Context window | 1M |
|---|---|
| Max output | 393K |
| Input modes | text |
| Tool use | yes |
| Reasoning | optional |
| Open weights | yes |
Provenance
| Price source | openrouter.ai |
|---|---|
| Fetched | 2026-11-22 |
| Quality data | verified |
| Cross-checked | vendor page |
Our take
editorial — not a measurementFor scheduled or cache-heavy work, confirm the selected seller’s billing rules before estimating savings. A hosted listing does not establish another endpoint’s time-of-day tariff. The accepted record lists open weights, but deployment still needs a licence and hardware review. Use the current price table and proof for supported measurements, and confirm the tariff with the endpoint you intend to use.
Strengths
- Open weights are recorded
- Text input, tool use and cached-input pricing are listed
Weaknesses
- Time-of-day terms must be established for the actual endpoint
- A missing licence detail needs investigation before self-hosting
Reach for it when
- Cache-heavy text workload evaluations
- Self-hosting assessments with a verified licence
Avoid it if
- Your savings calculation relies on an unverified schedule discount
- You require multimodal input
Sources: api-docs.deepseek.com openrouter.ai
DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with 1.6T total parameters and 49B activated parameters, supporting a 1M-token context window. It is designed for advanced reasoning, coding,...
Questions this page answers
What does DeepSeek V4 Pro 0423 cost?
$0.900/M in, $1.70/M out in this catalogue, as of the fetch date on this page. That is the API row, not a subscription.
Does DeepSeek V4 Pro 0423 have an independent quality score?
DeepSeek V4 Pro 0423 has an LMArena score in this catalogue.
What beats DeepSeek V4 Pro 0423?
GLM 5.3 has an equal-or-higher measured score and an equal-or-lower price.
Does this page use Artificial Analysis scores for DeepSeek V4 Pro 0423?
No. Artificial Analysis figures are not published here. Quality on this page is LMArena Elo where a score exists; otherwise the row is unrated.
Is the cheapest DeepSeek V4 Pro 0423 endpoint the same product?
Headline cheapest is a lower precision. Like-for-like at the best declared precision is $0.900/M from SiliconFlow.
MODEL MONUMENT