Compare / DeepSeek V3.2 vs GPT-4o (2024-08-06)
DeepSeek V3.2 vs GPT-4o (2024-08-06)
Which is cheaper, which is more capable, and which does better at each kind of work, from 15 results on 3 sources that measured both models.
DeepSeek V3.2 or GPT-4o (2024-08-06)? The short answer
- DeepSeek V3.2 is cheaper: $0.46 against $4.38 per million tokens, 9.4 times less.
- DeepSeek V3.2 scores higher on the Artificial Analysis Intelligence Index: 21.5 against 7.7.
- On our business benchmarks, DeepSeek V3.2 ranks higher for reasoning and knowledge, writing and sticking to the facts.
At a glance
| DeepSeek V3.2 | GPT-4o (2024-08-06) | |
|---|---|---|
| Price per million tokens | $0.30 in · $0.96 out | $2.50 in · $10.00 out |
| Blended (3 in : 1 out) | $0.46 | $4.38 |
| Cheapest host, blended | $0.23 (GMICloud, FP8) | $4.38 (Azure) |
| Intelligence Index | 21.5 | 7.7 |
| Spring Prompt overall | – | – |
| Output speed, tokens a second | – | – |
| Context window | 163,840 tokens | 128,000 tokens |
| Weights | Open | Closed |
| Developer based in | China | the US |
Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.
Which is better for what
Each model's rank among models available today, for each kind of work, from our own benchmarks and licensed sources. The better rank is in green.
| Use case | DeepSeek V3.2 | GPT-4o (2024-08-06) |
|---|---|---|
| Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls | 83 of 161 | not measured |
| Reasoning and knowledgeHard questions across science, maths and general knowledge | 81 of 195 | 176 of 195 |
| WritingWhat people prefer in blind comparisons, and judged writing quality | 64 of 187 | 152 of 187 |
| Sticking to the factsSummarising without inventing things, and factual answers | 56 of 153 | 105 of 153 |
Every shared result
| Benchmark | Metric | DeepSeek V3.2 | GPT-4o (2024-08-06) |
|---|---|---|---|
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,423 | 1,302 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,400 | 1,327 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,453 | 1,312 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,420 | 1,319 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,425 | 1,336 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,410 | 1,341 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 21.5 | 7.70 |
| Artificial Analysisreported | GPQA Diamond% of questions | 84.0% | 52.1% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 24.6% | 2.3% |
| Artificial Analysisreported | IFBench% of instructions | 60.7% | 36.0% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 73.3% | 41.0% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 35.6% | 8.3% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 90.6% | 28.9% |
| Vectara Hallucination Leaderboardreported | Answer rate% of documents | 92.6% | 93.8% |
| Vectara Hallucination Leaderboardreported | Hallucination rate% of summaries | 6.3% | 9.6% |
Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.
Quick answers
Is DeepSeek V3.2 better than GPT-4o (2024-08-06)?
It depends on the task. Among models available today, on our benchmarks and licensed sources, DeepSeek V3.2 ranks higher for reasoning and knowledge, writing and sticking to the facts.
Which is cheaper, DeepSeek V3.2 or GPT-4o (2024-08-06)?
DeepSeek V3.2 costs $0.46 per million tokens (three input to one output), against $4.38 for GPT-4o (2024-08-06).
Run DeepSeek V3.2 and GPT-4o (2024-08-06) on your own prompt
Benchmarks aren't your data. Try both side by side in the playground, with the cost of every answer.