DeepSeek-V3 vs GPT-5
A complete head-to-head: pricing, context window, benchmarks, modality coverage, and openness, with a programmatic verdict synthesized from the underlying data.
Updated
DeepSeek-V3 specs · GPT-5 specs- PriceDeepSeek-V3
DeepSeek-V3 is roughly 9.1× cheaper on output tokens ($1.10 vs $10.00 per 1M).
- Context windowGPT-5
GPT-5 accepts 400K tokens vs 128K, 3.1× the room for long documents and codebases.
- BenchmarksGPT-5
GPT-5 leads in 2 of 2 shared benchmarks; the biggest gap is on HumanEval (Python coding), where it scores 96.2 vs 90.0.
- ModalitiesGPT-5
GPT-5 supports 3 modalities (text, vision, audio) vs 1 for DeepSeek-V3.
- OpennessDeepSeek-V3
DeepSeek-V3 ships open weights (MIT); GPT-5 is API-only.
On balance GPT-5 edges ahead, winning 3 of 5 categories against DeepSeek-V3's 2. DeepSeek-V3 is roughly 9.1× cheaper on output tokens ($1.10 vs $10.00 per 1M). GPT-5 accepts 400K tokens vs 128K, 3.1× the room for long documents and codebases.
GPT-5 leads in 2 of 2 shared benchmarks; the biggest gap is on HumanEval (Python coding), where it scores 96.2 vs 90.0. They differ in modality coverage, DeepSeek-V3 handles text while GPT-5 handles text, vision, audio, which can be the deciding factor before you even look at benchmarks. DeepSeek-V3 ships open weights (MIT); GPT-5 is API-only.
GPT-5 is the newer of the two, released 7 months after DeepSeek-V3, which usually means a more recent knowledge cutoff and updated safety post-training. DeepSeek-V3 is usually picked for open source llm and commercial use llm workloads, while GPT-5 sees more deployments in coding llm and agents. If pricing matters more than every last benchmark point, run the numbers in the calculator below before committing.
Side-by-side specs
| Spec | DeepSeek-V3 | GPT-5 |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Released | Dec 2024 | Aug 2025 |
| Modalities | text | text, vision, audio |
| Context window | 128K tokens | 400K tokens |
| Max output | , | 128K tokens |
| Input · 1M | $0.27 / 1M tokens | $1.25 / 1M tokens |
| Output · 1M | $1.10 / 1M tokens | $10.00 / 1M tokens |
| Knowledge cutoff | 2024-07 | 2024-09 |
| Open weights | Yes (MIT) | No |
| API available | Yes | Yes |
Pricing at scale
What you'd actually pay at typical workloads. Numbers come from each model's published per-million-token rates.
- Light usage, 100k in / 50k out per day$2.46 vs $18.75
- Heavy usage, 1M in / 500k out per day$24.60 vs $188
- RAG workload, 5M in / 200k out per day$47.10 vs $248
Light usage, 100k in / 50k out per day: $2.46 vs $18.75 per month, model A comes out ahead. Heavy usage, 1M in / 500k out per day: $24.60 vs $188 per month, model A comes out ahead. RAG workload, 5M in / 200k out per day: $47.10 vs $248 per month, model A comes out ahead.
Estimated spend for the listed models at your usage. Numbers are derived from each model's published per-million-token rates.
- DeepSeek-V3$0.082
- GPT-5$0.625
Benchmarks compared
Only sourced numbers. Where a benchmark is missing for one model we show the available value rather than fabricating the other.
- MMLU88.591.4
- HumanEval90.096.2
- GPQA,89.4
- SWE-bench Verified,74.9
DeepSeek-V3 fits when…
- Frontier-level quality at open-weight prices
- MIT license, clean commercial use
- Cheap to serve via MoE architecture
- Cost-sensitive workloads, 9.1× cheaper than GPT-5 on output tokens.
- Self-hosting and on-prem requirements, open weights (MIT).
GPT-5 fits when…
- Unified model, reasoning routed automatically per query
- Excellent tool-use and JSON-mode discipline
- Strong agent performance on SWE-bench Verified
- Long-context tasks, handles 400K tokens vs 128K for DeepSeek-V3.
- Multimodal needs covering vision, audio.
Consider DeepSeek-R1
First open-weight reasoning model to match o1, the release that proved RL-from-scratch reasoning training was reproducible.
Frequently asked
Is DeepSeek-V3 or GPT-5 cheaper?
DeepSeek-V3 is cheaper at $1.10 / 1M tokens per million output tokens, vs $10.00 / 1M tokens for GPT-5.Which has the larger context window?
GPT-5 accepts 400K tokens vs 128K for DeepSeek-V3.Is DeepSeek-V3 or GPT-5 better for coding?
Both DeepSeek-V3 and GPT-5 are competitive on coding benchmarks. See each model's individual spec page for HumanEval and SWE-bench scores where published. For an opinionated pick, consult our Best LLM for Coding ranking.Are either of these models open source?
DeepSeek-V3 ships open weights (MIT). GPT-5 is API-only.When were DeepSeek-V3 and GPT-5 released?
DeepSeek-V3 was released by DeepSeek on 2024-12-26. GPT-5 was released by OpenAI on 2025-08-07.
The week's AI launches, in your inbox.
One short email every Friday, new models, leaks, and quietly-shipped APIs you missed.