Cheaper overall
Gemini 3.8 Flash
24% lower blended rate
Higher benchmark scores
Gemini 3.8 Flash
leads on 9 of 11 figures
Cheaper cached input
Deepseek v4 Pro-0813
$0.044 vs $0.075, 41% less
Where each one wins
Gemini 3.8 Flash
- Cheaper input tokens $0.75 vs $1.32, 43% less
- Cheaper output tokens $3.75 vs $3.96, 5% less
- Cheaper on all four workloads
- Higher overall intelligence (Intelligence Index) 40.9 vs 36
- Higher coding ability (Coding Index) 76.3 vs 68.8
- Higher graduate-level science (GPQA Diamond) 95.3 vs 92.8
- Ahead on 6 more benchmark figures
Deepseek v4 Pro-0813
- Cheaper cached input $0.044 vs $0.075, 41% less
- Longer maximum output 384,000 tokens
- Higher multi-step tool use (Agentic Index) 41.3 vs 40.2
- Higher real-world task performance (GDPval) 47.9 vs 47.3
What four workloads cost
One run, and the same run a thousand times.
Gemini 3.8 Flash is cheaper on all four workloads, by much the same margin each time — about 1.4 times, from $0.0026 against $0.0033 on a chat turn to $0.3113 against $0.5399 on a long document.
| Workload | Gemini 3.8 Flash | Deepseek v4 Pro-0813 | ×1,000 | Difference |
|---|---|---|---|---|
| Chat turn 1,000 in / 500 out | $0.0026 | $0.0033 | $2.63 / $3.30 | Gemini 3.8 Flash is 20% cheaper |
| RAG answer 10,000 in / 800 out | $0.0105 | $0.0164 | $10.50 / $16.37 | Gemini 3.8 Flash is 36% cheaper |
| Agent loop 32,000 in / 700 out × 20 calls | $0.1275 | $0.1346 | $127.50 / $134.64 | Gemini 3.8 Flash is 5% cheaper |
| Long document 400,000 in / 3,000 out | $0.3113 | $0.5399 | $311.25 / $539.88 | Gemini 3.8 Flash is 42% cheaper |
Price per 1M tokens
Neither model changes its rate with prompt length, so these rates apply to every request.
| Context band | Price | Gemini 3.8 Flash | Deepseek v4 Pro-0813 | Difference |
|---|---|---|---|---|
| Any prompt length | Input | $0.75 | $1.32 | Gemini 3.8 Flash is 43% cheaper |
| Any prompt length | Cached input | $0.075 | $0.044 | Deepseek v4 Pro-0813 is 41% cheaper |
| Any prompt length | Output | $3.75 | $3.96 | Gemini 3.8 Flash is 5% cheaper |
Blended: Gemini 3.8 Flash $1.50, Deepseek v4 Pro-0813 $1.98 per 1M — one rate at 3:1 input to output, for comparing two models at a glance.
Both priced by Perplexity. Standard tier, pay-as-you-go.
Benchmarks
They share 11 figures: Gemini 3.8 Flash leads on 9, Deepseek v4 Pro-0813 on 2. The widest gap is 42.9 points, on answer reliability (Omniscience Non-Hallucination).
Overall intelligence
Intelligence Index — Overall intelligence across reasoning, knowledge & math evals
Gemini 3.8 Flash ahead by 4.9
Coding ability
Coding Index — Coding ability across software-engineering evals
Gemini 3.8 Flash ahead by 7.5
Multi-step tool use
Agentic Index — Tool use & multi-step agent task performance
Deepseek v4 Pro-0813 ahead by 1.1
Graduate-level science
GPQA Diamond — Graduate-level questions in biology, chemistry & physics
Gemini 3.8 Flash ahead by 2.5
Expert-exam performance
Humanity's Last Exam — Expert-level questions across many academic domains
Gemini 3.8 Flash ahead by 6.8
Long-context reasoning
AA-LCR — Long-context reasoning across large inputs
Gemini 3.8 Flash ahead by 3.7
Scientific coding
SciCode — Research-level scientific coding tasks
Gemini 3.8 Flash ahead by 5.6
Physics research reasoning
CritPt — Unpublished physics research reasoning problems
Gemini 3.8 Flash ahead by 0.3
Real-world task performance
GDPval — Economically valuable, real-world knowledge work
Deepseek v4 Pro-0813 ahead by 0.6
Factual accuracy
Omniscience Accuracy — Breadth of factual knowledge across domains
Gemini 3.8 Flash ahead by 5.5
Answer reliability
Omniscience Non-Hallucination — How reliably the model avoids fabricated answers
Gemini 3.8 Flash ahead by 42.9
Each figure is that model's best published run, with the effort level named beside it — where these scores come from.
Specifications
Gemini 3.8 Flash also takes audio, file, image and video. Only Gemini 3.8 Flash lists computer use; only Deepseek v4 Pro-0813 lists open weights.
| Specification | Gemini 3.8 Flash | Deepseek v4 Pro-0813 |
|---|---|---|
| Context window | 1,048,576 tokens | 1,048,576 tokens |
| Max output | 65,536 tokens | 384,000 tokens |
| Takes in | Text, audio, image, video, file | Text |
| Puts out | Text | Text |
| Released | 2 Sep 2026 | 12 Aug 2026 |
| Status | Active | Active |
| Sold by | Perplexity | Perplexity |
| Tool use | Yes | Yes |
| Structured outputs | Yes | Yes |
| Web search | Not listed | Not listed |
| Prompt caching | Yes | Yes |
| Code execution | Yes | Yes |
| Computer use | Yes | Not listed |
| Open weights | Not listed | Yes |
| Reasoning | Yes | Yes |
FAQs
- Is Gemini 3.8 Flash cheaper than Deepseek v4 Pro-0813?
- Yes. A 1,000-token prompt with a 500-token reply costs $0.0026 on Gemini 3.8 Flash and $0.0033 on Deepseek v4 Pro-0813, and the same model is cheaper on every workload on this page.
- How much do Gemini 3.8 Flash and Deepseek v4 Pro-0813 cost per 1M tokens?
- Gemini 3.8 Flash costs $0.75 for input and $3.75 for output. Deepseek v4 Pro-0813 costs $1.32 and $3.96. Standard pay-as-you-go rates, as of 17 Oct 2026.
- Which is better, Gemini 3.8 Flash or Deepseek v4 Pro-0813?
- On published benchmarks, Gemini 3.8 Flash. It leads on 9 of the 11 figures both models report, including overall intelligence (Intelligence Index), where it scores 40.9 against 36.