Cheaper overall
Deepseek v4 Pro-0813
65% lower blended rate
Higher benchmark scores
GPT-5.4
leads on 7 of 9 figures
Bigger context window
GPT-5.4
1,050,000 vs 1,048,576 tokens
Cheaper cached input
Deepseek v4 Pro-0813
$0.044 vs $0.25, 82% less
Where each one wins
Deepseek v4 Pro-0813
- Cheaper input tokens $1.32 vs $2.50, 47% less
- Cheaper output tokens $3.96 vs $15.00, 74% less
- Cheaper cached input $0.044 vs $0.25, 82% less
- Cheaper on all four workloads
- Longer maximum output 384,000 tokens
- Higher graduate-level science (GPQA Diamond) 92.8 vs 92
- Higher real-world task performance (GDPval) 47.9 vs 37.4
GPT-5.4
- Larger context window 1,050,000 tokens
- Higher overall intelligence (Intelligence Index) 39 vs 36
- Higher coding ability (Coding Index) 71.1 vs 68.8
- Higher expert-exam performance (Humanity's Last Exam) 43.7 vs 41
- Ahead on 4 more benchmark figures
- Offers Batch, Flex and Priority pricing not listed for Deepseek v4 Pro-0813
What four workloads cost
One run, and the same run a thousand times.
Deepseek v4 Pro-0813 is cheaper on all four workloads, by much the same margin each time — about 3.1 times, from $0.0033 against $0.0100 on a chat turn to $0.5399 against $2.07 on a long document.
| Workload | Deepseek v4 Pro-0813 | GPT-5.4 | ×1,000 | Difference |
|---|---|---|---|---|
| Chat turn 1,000 in / 500 out | $0.0033 | $0.0100 | $3.30 / $10.00 | Deepseek v4 Pro-0813 is 67% cheaper |
| RAG answer 10,000 in / 800 out | $0.0164 | $0.0370 | $16.37 / $37.00 | Deepseek v4 Pro-0813 is 56% cheaper |
| Agent loop 32,000 in / 700 out × 20 calls | $0.1346 | $0.4600 | $134.64 / $460.00 | Deepseek v4 Pro-0813 is 71% cheaper |
| Long document 400,000 in / 3,000 out | $0.5399 | $2.07 | $539.88 / $2,067.50 | Deepseek v4 Pro-0813 is 74% cheaper |
Price per 1M tokens
Above 272,000 input tokens the rates change, and the new rate applies to the whole request, not just the tokens past that point: GPT-5.4's input goes from $2.50 to $5.00. It does not change which of the two is cheaper.
| Context band | Price | Deepseek v4 Pro-0813 | GPT-5.4 | Difference |
|---|---|---|---|---|
| Prompts up to 272,000 tokens | Input | $1.32 | $2.50 | Deepseek v4 Pro-0813 is 47% cheaper |
| Prompts up to 272,000 tokens | Cached input | $0.044 | $0.25 | Deepseek v4 Pro-0813 is 82% cheaper |
| Prompts up to 272,000 tokens | Output | $3.96 | $15.00 | Deepseek v4 Pro-0813 is 74% cheaper |
| Prompts over 272,000 tokens | Input | $1.32 | $5.00 | Deepseek v4 Pro-0813 is 74% cheaper |
| Prompts over 272,000 tokens | Cached input | $0.044 | $0.50 | Deepseek v4 Pro-0813 is 91% cheaper |
| Prompts over 272,000 tokens | Output | $3.96 | $22.50 | Deepseek v4 Pro-0813 is 82% cheaper |
Blended: Deepseek v4 Pro-0813 $1.98, GPT-5.4 $5.625 per 1M — one rate at 3:1 input to output, for comparing two models at a glance.
Deepseek v4 Pro-0813 priced by Perplexity, GPT-5.4 by OpenAI. Standard tier, pay-as-you-go.
Other pricing tiers
Only GPT-5.4 lists Batch, Flex and Priority, at up to 50% off its own standard rate; Deepseek v4 Pro-0813 offers none of them.
Every tier either model sells is here, so this is the whole price sheet. A saving is measured against that model's own standard rate — how we read tiers.
| Tier | Deepseek v4 Pro-0813 | GPT-5.4 | Cheaper |
|---|---|---|---|
| Batch only on GPT-5.4 | Not offered | $1.25 in / $7.50 out 50% off Standard | — |
| Flex only on GPT-5.4 | Not offered | $1.25 in / $7.50 out 50% off Standard | — |
| Priority only on GPT-5.4 | Not offered | $5.00 in / $30.00 out | — |
Benchmarks
They share 9 figures: Deepseek v4 Pro-0813 leads on 2, GPT-5.4 on 7. The widest gap is 12.2 points, on answer reliability (Omniscience Non-Hallucination).
Overall intelligence
Intelligence Index — Overall intelligence across reasoning, knowledge & math evals
GPT-5.4 ahead by 3
Coding ability
Coding Index — Coding ability across software-engineering evals
GPT-5.4 ahead by 2.3
Graduate-level science
GPQA Diamond — Graduate-level questions in biology, chemistry & physics
Deepseek v4 Pro-0813 ahead by 0.8
Expert-exam performance
Humanity's Last Exam — Expert-level questions across many academic domains
GPT-5.4 ahead by 2.7
Long-context reasoning
AA-LCR — Long-context reasoning across large inputs
GPT-5.4 ahead by 1.7
Physics research reasoning
CritPt — Unpublished physics research reasoning problems
GPT-5.4 ahead by 5.4
Real-world task performance
GDPval — Economically valuable, real-world knowledge work
Deepseek v4 Pro-0813 ahead by 10.5
Factual accuracy
Omniscience Accuracy — Breadth of factual knowledge across domains
GPT-5.4 ahead by 1.7
Answer reliability
Omniscience Non-Hallucination — How reliably the model avoids fabricated answers
GPT-5.4 ahead by 12.2
Each figure is that model's best published run, with the effort level named beside it — where these scores come from.
Specifications
GPT-5.4 holds 1,424 more input tokens in one request. GPT-5.4 also takes file and image.
| Specification | Deepseek v4 Pro-0813 | GPT-5.4 |
|---|---|---|
| Context window | 1,048,576 tokens | 1,050,000 tokens |
| Max output | 384,000 tokens | 128,000 tokens |
| Takes in | Text | Text, image, file |
| Puts out | Text | Text |
| Knowledge cutoff | — | 31 Aug 2025 |
| Released | 12 Aug 2026 | 5 Mar 2026 |
| Status | Active | Active |
| Sold by | Perplexity | OpenAI |
FAQs
- Is Deepseek v4 Pro-0813 cheaper than GPT-5.4?
- Yes. A 1,000-token prompt with a 500-token reply costs $0.0033 on Deepseek v4 Pro-0813 and $0.0100 on GPT-5.4, and the same model is cheaper on every workload on this page.
- How much do Deepseek v4 Pro-0813 and GPT-5.4 cost per 1M tokens?
- Deepseek v4 Pro-0813 costs $1.32 for input and $3.96 for output. GPT-5.4 costs $2.50 and $15.00. Standard pay-as-you-go rates, as of 17 Oct 2026.
- Which is better, Deepseek v4 Pro-0813 or GPT-5.4?
- On published benchmarks, GPT-5.4. It leads on 7 of the 9 figures both models report, including overall intelligence (Intelligence Index), where it scores 39 against 36.
- Does Deepseek v4 Pro-0813 or GPT-5.4 offer batch pricing?
- GPT-5.4 only. Its batch input costs $1.25, 50% off its standard rate, and Deepseek v4 Pro-0813 lists no batch tier at all.
- Does Deepseek v4 Pro-0813 or GPT-5.4 have a bigger context window?
- GPT-5.4, at 1,050,000 tokens against 1,048,576. That is 1,424 more input tokens in a single request.