Azure AI Foundry vs Azure Cognitive Services.
Prices checked
Azure AI Foundry is cheaper on 3 of the 57 models both sell, the widest gap GPT-5.6 Luna at $1.80 against $9 per M tokens, checked .
Where to start
Azure AI Foundryall 69 models →Sells OpenAI's and partners' models under Azure terms, priced per region.
Cheapest of the shared models
6¢/M tokens
text-embedding-3-small
Azure Cognitive Servicesall 59 models →Many makers' models, one API and one bill.
Cheapest of the shared models
6¢/M tokens
text-embedding-3-small
Side by side
| Fact | Azure AI Foundry | Azure Cognitive Services |
|---|---|---|
| Kind of provider | Cloud platform | Router |
| Models we track | 69 | 59 |
| Cheaper of the two | 3 of 57 | 1 of 57 |
| Trains on your prompts | does not say | does not say |
| Keeps your requests | does not say | does not say |
| How you pay | does not say | does not say |
| Smallest top-up | does not say | does not say |
| Free tier | none listed | none listed |
| Where it runs | does not say | does not say |
Every model both sell
57 models, most widely sold first. Per M tokens, blended 3 in : 1 out.
| Model | Azure AI Foundry | Azure Cognitive Services |
|---|---|---|
| Kimi K2.6Moonshot AI · 262k context | $6.85 $0.95 in · $4 out | $6.85 $0.95 in · $4 out |
| DeepSeek V3.2DeepSeek · 164k context | $3.42 $0.58 in · $1.68 out | $3.42 $0.58 in · $1.68 out |
| Claude Sonnet 4.6Anthropic · 1M context | $24 $3 in · $15 out | $24 $3 in · $15 out |
| Claude Haiku 4.5 (latest)Anthropic · 200k context | $8 $1 in · $5 out | $8 $1 in · $5 out |
| Claude Opus 4.8Anthropic · 1M context | $40 $5 in · $25 out | $40 $5 in · $25 out |
| Claude Opus 4.7Anthropic · 1M context | $40 $5 in · $25 out | $40 $5 in · $25 out |
| GPT-5.5OpenAI · 1.1M context | $45 $5 in · $30 out | $45 $5 in · $30 out |
| Kimi K2.5Moonshot AI · 262k context | $4.80 $0.60 in · $3 out | $4.80 $0.60 in · $3 out |
| GPT-5.4OpenAI · 1.1M context | $22.50 $2.50 in · $15 out | $22.50 $2.50 in · $15 out |
| Llama-3.3-70B-InstructMeta · 128k context | $2.84 $0.71 in · $0.71 out | $2.84 $0.71 in · $0.71 out |
| Claude Fable 5Anthropic · 1M context | $80 $10 in · $50 out | $80 $10 in · $50 out |
| Claude Sonnet 5Anthropic · 1M context | $16 $2 in · $10 out | $16 $2 in · $10 out |
| GPT-5.6 SolOpenAI · 1.1M context | $32 $4 in · $20 out | $45 $5 in · $30 out |
| Claude Opus 4.6Anthropic · 1M context | $40 $5 in · $25 out | $40 $5 in · $25 out |
| GPT-5.6 LunaOpenAI · 1.1M context | $1.80 $0.20 in · $1.20 out | $9 $1 in · $6 out |
| Claude Opus 5Anthropic · 1M context | $40 $5 in · $25 out | $40 $5 in · $25 out |
| GPT-5.6 TerraOpenAI · 1.1M context | $18 $2 in · $12 out | $22.50 $2.50 in · $15 out |
| Claude Sonnet 4.5 (latest)Anthropic · 1M context | $24 $3 in · $15 out | $24 $3 in · $15 out |
| Claude Opus 4.5 (latest)Anthropic · 200k context | $40 $5 in · $25 out | $40 $5 in · $25 out |
| GPT-5.4 miniOpenAI · 400k context | $6.75 $0.75 in · $4.50 out | $6.75 $0.75 in · $4.50 out |
| R1 0528DeepSeek · 164k context | $9.45 $1.35 in · $5.40 out | $9.45 $1.35 in · $5.40 out |
| Claude Opus 5.5Anthropic · 1M context | $32 $4 in · $20 out | $32 $4 in · $20 out |
| GPT-5 MiniOpenAI · 400k context | $2.75 $0.25 in · $2 out | $2.75 $0.25 in · $2 out |
| GPT-5.4 nanoOpenAI · 400k context | $1.85 $0.20 in · $1.25 out | $1.85 $0.20 in · $1.25 out |
| Claude Fable 5.1Anthropic · 1M context | $80 $10 in · $50 out | $80 $10 in · $50 out |
| GPT-5OpenAI · 400k context | $13.75 $1.25 in · $10 out | $13.75 $1.25 in · $10 out |
| GPT-5.1OpenAI · 400k context | $13.75 $1.25 in · $10 out | $13.75 $1.25 in · $10 out |
| GPT-5.2 ChatOpenAI · 128k context | $19.25 $1.75 in · $14 out | $19.25 $1.75 in · $14 out |
| GPT-6 AstraOpenAI · 1.1M context | $80 $10 in · $50 out | $80 $10 in · $50 out |
| GPT-4.1 miniOpenAI · 1M context | $2.80 $0.40 in · $1.60 out | $2.80 $0.40 in · $1.60 out |
| GPT-5 NanoOpenAI · 400k context | $0.55 5¢ in · $0.40 out | $0.55 5¢ in · $0.40 out |
| GPT-5.3 CodexOpenAI · 400k context | $19.25 $1.75 in · $14 out | $19.25 $1.75 in · $14 out |
| GPT-6 LunaOpenAI · 1.1M context | $0.80 $0.10 in · $0.50 out | $0.80 $0.10 in · $0.50 out |
| GPT-6 SolOpenAI · 1.1M context | $16 $2 in · $10 out | $16 $2 in · $10 out |
| GPT-4.1OpenAI · 1M context | $14 $2 in · $8 out | $14 $2 in · $8 out |
| GPT-4o miniOpenAI · 128k context | $1.05 $0.15 in · $0.60 out | $1.05 $0.15 in · $0.60 out |
| GPT-4.1 nanoOpenAI · 1M context | $0.70 $0.10 in · $0.40 out | $0.70 $0.10 in · $0.40 out |
| Claude Sonnet 5.5Anthropic · 1M context | $16 $2 in · $10 out | $16 $2 in · $10 out |
| o3OpenAI · 200k context | $14 $2 in · $8 out | $14 $2 in · $8 out |
| o4-miniOpenAI · 200k context | $7.70 $1.10 in · $4.40 out | $7.70 $1.10 in · $4.40 out |
The 40 most widely sold of 57 · every row in the open dataset
How we count
Azure AI Foundry is cheaper on 3 of the 57 models both sell, Azure Cognitive Services on 1, and 53 cost the same at both. Azure AI Foundry is a cloud platform and Azure Cognitive Services a router, so the two catalogs overlap without matching.
Each price is the cheapest that provider charges for the model, per M tokens, blended 3 in : 1 out. The smaller line under it is the price in and the price out.
A free tier is a limit rather than a price, and a price below the maker's own with nothing published to account for it is left out, so neither can be marked as the cheaper of the two.
Frequently asked
- Is Azure AI Foundry cheaper than Azure Cognitive Services?
- On the 57 models both sell, Azure AI Foundry is cheaper on 3, Azure Cognitive Services on 1, and 53 cost the same at both. The widest gap is GPT-5.6 Luna: $1.80 per million tokens against $9. Prices are per M tokens, blended 3 in : 1 out, and read on 29 Sept 2026.
- Which models do Azure AI Foundry and Azure Cognitive Services both sell?
- 57 of the models we track, led by Kimi K2.6, DeepSeek V3.2, Claude Sonnet 4.6. Azure AI Foundry sells 69 in all and Azure Cognitive Services 59, so most of each catalog has no counterpart at the other.
- Does Azure AI Foundry or Azure Cognitive Services train on what you send it?
- Azure AI Foundry does not say whether it trains on what you send it, and does not say how long it keeps requests. Azure Cognitive Services does not say whether it trains on what you send it, and does not say how long it keeps requests. A provider that may train on prompts is often how a price below the maker's own is paid for.
- How current are these prices?
- Both catalogs are read every day. Azure AI Foundry was last read 29 Sept 2026 and Azure Cognitive Services 29 Sept 2026; the whole index was refreshed 29 Sept 2026. Prices exclude tax, and a price below the maker's own with nothing published to account for it is left out of this table entirely.
More comparisons