Provider comparison · Updated 10 Oct 2026
DeepSeek vs Microsoft
13 models vs 9 models
DeepSeek is cheaper at the median: $0.27 against $0.35 per 1M input tokens across paid models, about 1.3× less. Its cheapest paid model, DeepSeek V4 Flash 0731 ($0.08), compares with Phi 4 ($0.07) at Microsoft. DeepSeek lists 13 models; Microsoft lists 9 models.
Ask an AI about this pageGChatGPTCClaudePPerplexityGeGeminiXGrok
Cite
Cite this page
ModelsAtlas. “DeepSeek vs Microsoft”. Retrieved 10 Oct 2026. https://modelsatlas.com/compare-providers/deepseek/microsoft
At a glance
Standard tier, paid models · highlighted = better
DeepSeek
Microsoft
Models listed
13
9
Cheapest input / 1M
$0.08
$0.07
Median in / out
$0.27 / $0.95
$0.35 / $0.38
Highest input / 1M
$0.70
$0.62
Largest context
1.05M
66K
Batch-tier models
1 of 13
0 of 9
Open-weight models
3
7
Free models
0
0
Vision models listed
1
1
Costs for real workloads
Cheapest paid model from each provider that fits the job · monthly
Derived from standard-tier prices, 30 days. “Fits the job” means the context window holds the prompt and reply. Excludes batch and caching.
Cheapest model for each capability
Input / output per 1M
| Capability | DeepSeek | Microsoft |
|---|---|---|
| 1M context | DeepSeek V4 Flash 0731 $0.08 / $0.17 | None listed |
Line-up by price tier
Paid models · input per 1M
| Tier | DeepSeek | Microsoft |
|---|---|---|
| Flagship $3+ | None listed | None listed |
| Mid $0.50–$3 | R1 0528 $0.70 | WizardLM-2 8x22B $0.62 |
| Budget under $0.50 | DeepSeek V4 Flash 0731 $0.08 DeepSeek V4 Flash Vision Exp $0.22 DeepSeek V3.1 $0.25 DeepSeek V3.2 $0.26 DeepSeek V3.1 Terminus $0.27 +3 more | Phi 4 $0.07 |
| Open weights any price | None listed | None listed |
Questions
Is DeepSeek or Microsoft cheaper?
DeepSeek at the median: $0.27 against $0.35 for Microsoft per 1M input tokens. Median is across each provider's paid standard-tier models, so a provider with many small models can look cheaper than its flagship is.
Which provider has more models?
DeepSeek lists more: DeepSeek 13, Microsoft 9. Counts are models with a page in our catalogue.
Do both offer batch pricing?
DeepSeek: 1 model list a batch tier. Microsoft: no batch tier listed. Where a model lists both, batch is a median 37% of standard for DeepSeek.
Which has the larger context window?
DeepSeek has the larger: DeepSeek V4 Flash 0731 (1.05M tokens) at DeepSeek, WizardLM-2 8x22B (66K) at Microsoft.
Does either provider offer open-weight models?
DeepSeek: 3 open-weight models. Microsoft: 7 open-weight models.
Does either have free models?
DeepSeek: none listed at no per-token charge. Microsoft: none.
Browse more providers
Keep going in your AI
Opens with this page as the source. We add nothing else to the prompt.
- Is DeepSeek or Microsoft cheaper?
- Which provider has more models?
- Do both offer batch pricing?