Phi 4 API pricing
Microsoft · Budget price tier · MIT · released 2025-01. Prices in USD per 1 million tokens. Data snapshot: 2026-11-12.
A support ticket workload of 100K requests a month costs $17.50 on Phi 4; the next cheaper model in the Budget price tier for this workload, Gemma 4 26B A4B, does it for $17.00. Phi 4 has held the same list price since 2026-08-15 and is the 22nd cheapest of 160 models in the Budget price tier. Only one host, DeepInfra, sells it, so there is no second price to shop. It is rated a stable release but does not meet all four criteria of the FinOps Friendly badge.
- Input: $0.07 per 1M tokens
- Output: $0.14 per 1M tokens
- Batch API (in/out): Not published
- Prompt-cache read: Not published
- Context window: 16K tokens
- Arena ELO: 1217 as of 2026-11-08
Market position
On blended price ($0.119 per 1M tokens at 30% input, 70% output) it is cheaper than 87% of models in the Budget price tier: 22nd cheapest of 160. Across every price tier it is cheaper than 93% of priced text models in the catalogue: 22nd cheapest of 303. Its Arena ELO of 1217 ranks 154th of the 160 models that carry one, above 4% of them. With an Arena score of 1217, Phi 4 sits below OptimNow's lowest working quality threshold, Standard (Arena 1420+), so the quality-bar method would not shortlist it for any of the 8 use cases the table prices. Each threshold is a distance below the best ranked model, which scores 1530 on Arena scores as of 8 Nov 2026.
Licence, size and capabilities
Phi 4 is open source: its weights are published under MIT, an OSI-approved licence with no field-of-use limit, so it can be self-hosted as well as bought by the token. Its published size is 14B parameters. Listed capabilities: Text. Filling its 16K context window once costs $0.00112 in input tokens at list price.
Pricing by host
A host is a company that runs the model and sells it by the token: the model maker, a cloud platform like Amazon Bedrock, Azure or Vertex, or an independent host. One model, several hosts, and often several prices. This model is sold by 1 host. The price we quote elsewhere on this page is marked. We quote one host's rate so that models stay comparable; it is not a recommendation to buy there.
DeepInfra is the only host with a published rate, $0.07 input / $0.14 output per 1M tokens, which is the price we quote.
| Host | Input $/1M | Output $/1M | Precision |
|---|---|---|---|
| DeepInfra (price we quote) | $0.07 | $0.14 | bf16 |
List prices as published on 2026-11-12, via OpenRouter. Negotiated rates, committed-use discounts and provisioned capacity are not shown.
Monthly cost at 100K requests, by business use case
From the after-discounts figure (prompt caching on the cached input share, batch API for async workloads, only where Microsoft publishes those rates), which assumes you implement both, to the list price. One figure means no discount is published.
| Use case | Tokens (in/out) | Monthly (after discounts – list) |
|---|---|---|
| Support Ticket | 1,500 / 500 | $18 |
| Knowledge Q&A | 2,000 / 800 | $25 |
| Meeting Summary | 10,000 / 1,200 | $87 |
| Marketing Content | 2,500 / 1,800 | $43 |
| Coding Task | 3,000 / 2,000 | $49 |
| Invoice Processing | 1,500 / 600 | $19 |
| Call Summary | 2,000 / 700 | $24 |
| Agent Workflow | 6,000 / 3,000 | $84 |
Price history
Phi 4 has been in our daily price archive since tracking began on 2026-08-15, when it listed at $0.07 input / $0.14 output per 1M tokens. No list price change since tracking began on 2026-08-15. Last checked against the 2026-11-12 snapshot.
Alternatives to Phi 4
1 of the 4 comparable models below costs less than Phi 4: Nova Micro 1.0 undercuts it by 9%, for 9 Arena points less.
- Next cheaper model in the Budget price tier: Command R7B (12-2024) (Cohere) at $0.1162 blended, 2% less.
- Next more expensive: Gemma 3 12B (Google) at $0.12 blended, 1% more.
The nearest models in the Budget price tier on Arena ELO, each with its blended price and score difference against Phi 4.
- Nova Micro 1.0 (Amazon): $0.035 in / $0.14 out per 1M, Arena ELO 1208 (9% cheaper, 9 ELO points lower)
- Nova Lite 1.0 (Amazon): $0.06 in / $0.24 out per 1M, Arena ELO 1229 (56% more expensive, 12 ELO points higher)
- Qwen2.5 Coder 32B Instruct (Alibaba): $0.66 in / $1.00 out per 1M, Arena ELO 1230 (655% more expensive, 13 ELO points higher)
- Gemma 2 27B (Google): $0.65 in / $0.65 out per 1M, Arena ELO 1231 (446% more expensive, 14 ELO points higher)
- Cheapest within 15 Arena points: Nova Micro 1.0 (Amazon), 9% cheaper, 9 ELO points lower, out of 6 models in that band.
- Highest Arena ELO within 20% of this price: Gemma 3 12B (Google), under 1% more expensive, 117 ELO points higher, out of 5 models in that price band.
Common questions about Phi 4 pricing
- How much does Phi 4 cost?
- Phi 4 costs $0.07 per 1M input tokens and $0.14 per 1M output tokens at DeepInfra, the host whose rate we quote. Prices as listed on 2026-11-12.
- Which host is cheapest for Phi 4?
- DeepInfra is the only host with a published rate, $0.07 input / $0.14 output per 1M tokens, which is the price we quote. Host rates as listed on 2026-11-12.
- Does Phi 4 offer batch or cached pricing?
- No. Neither a batch rate nor a prompt-cache rate is published for Phi 4 as of 2026-11-12.
- What would Phi 4 cost per month?
- With the Support Ticket profile (1,500 input and 500 output tokens per request), 100,000 requests a month cost about $18 on Phi 4 at the list price quoted on 2026-11-12.
- Can Phi 4 be self-hosted?
- Yes. Phi 4 is open source: its weights are published under MIT, an OSI-approved licence with no field-of-use limit, so it can be self-hosted as well as bought by the token.
- Has the price of Phi 4 changed?
- Phi 4 has been in our daily price archive since tracking began on 2026-08-15, when it listed at $0.07 input / $0.14 output per 1M tokens. No list price change since tracking began on 2026-08-15. Last checked against the 2026-11-12 snapshot.
Its weights are published, which changes what you may run yourself: the cost gap between open weights and proprietary APIs.
These are list prices. Your real cost also depends on caching and batching, negotiated discounts, marketplace agreements, and the contracts you already have. OptimNow audits exactly that.
Build the business case · Get a FinOps review
LLM pricing · Models by maker · Cloud compute pricing · Compare models · Price barometer · FinOps guides · API and MCP server