Phi 4 with Jev
Phi 4 is a model from Microsoft, released 2025-01-10. It costs $0.070 per million input tokens and $0.140 per million output tokens, reads up to 16K tokens of context, and writes up to 14,745 tokens in one reply. Through llm11, set model to microsoft/phi-4.
- Input price
- $0.070 / 1M
- Output price
- $0.140 / 1M
- Context window
- 16K tokens
- Max output
- 15K tokens
- Released
- 2025-01-10
- Accepts
- text
Read from the live catalogue and refreshed daily. At blended list price, 92% of the 332 priced models we can call cost more.
Using Phi 4 with Jev
Jev is TypeSafe AI's System One model. It reads each request first and decides which model in your pool answers, with a calibrated confidence on that call. Phi 4 does the answering when Jev sends it work, or when you name it yourself. At list price Phi 4 falls in the llm11-fast band, but no pack routes to it because its context window is under 32K tokens, the floor for pack members. You can still pin it: pass microsoft/phi-4 as model and routing steps aside while verification still runs.
Jev is new, so we have not published results for Phi 4 with and without it, and this page does not invent any. Every response comes with a receipt that names the model that answered and what it cost against your most expensive model, so you can measure the pairing on your own traffic.
What a request costs
List price arithmetic, the same sum a receipt does. llm11 adds nothing per request; the only fee is on buying credits.
| Request | Tokens in / out | One | A thousand |
|---|---|---|---|
| Short chat turn | 500 / 200 | $0.00006 | $0.063 |
| Question over retrieved documents | 4,000 / 500 | $0.00035 | $0.350 |
| Long document summary | 30,000 / 1,000 | $0.00224 | $2.24 |
Cheaper models to try
The catalogue has no benchmark score for Phi 4, so this is a price ordering. It does not claim these answer as well.
| Model | Input / 1M | Output / 1M | Context | Released |
|---|---|---|---|---|
| Qwen Qwen3 30B A3B Instruct 2507 | $0.048 | $0.193 | 262K | |
| Google Gemma 3 12B | $0.050 | $0.150 | 131K | |
| OpenAI gpt-oss-120b | $0.037 | $0.170 | 131K | |
| Cohere Command R7B (12-2024) | $0.037 | $0.150 | 128K |
Call it
Same OpenAI request shape. Naming the model pins it, so nothing is routed, and the answer is still checked.
python
from openai import OpenAI
client = OpenAI(base_url="https://www.llm11.com/v1", api_key="llm11_live_...")
res = client.chat.completions.create(
model="microsoft/phi-4",
messages=[{"role": "user", "content": "Hello"}],
)More from Microsoft
| Model | Input / 1M | Output / 1M | Context | Released |
|---|---|---|---|---|
| WizardLM-2 8x22B | $0.620 | $0.620 | 66K |
Questions
- What is Phi 4 with Jev?
- Jev is TypeSafe AI's System One model. It reads each request first and decides which model in your pool answers, with a calibrated confidence on that call. Phi 4 does the answering when Jev sends it work, or when you name it yourself. At list price Phi 4 falls in the llm11-fast band, but no pack routes to it because its context window is under 32K tokens, the floor for pack members. You can still pin it: pass microsoft/phi-4 as model and routing steps aside while verification still runs. Routing decisions are made by Jev, TypeSafe AI's System One model.
- Is Phi 4 a System One model?
- No. Phi 4 answers requests. The System One model is Jev, which decides which model answers each one, so the two do different jobs and llm11 uses both.
- How much does Phi 4 cost?
- $0.070 per million input tokens and $0.140 per million output tokens. A 4,000 token prompt with a 500 token answer costs about $0.00035, so a thousand of them is about $0.350. llm11 passes provider list price through and charges 5% when you buy credits.
- What is the Phi 4 context window?
- 16,384 tokens, with up to 14,745 tokens in a single reply.
- Does Phi 4 support tool calling and structured output?
- The catalogue lists structured output, but it does not list tool calling and reasoning controls.
- Can I use Phi 4 through llm11?
- Yes. Set model to microsoft/phi-4 on the OpenAI-compatible endpoint and the request goes to it directly, with verification still running.
- Does Jev route to Phi 4?
- Routing decisions are made by Jev, TypeSafe AI's System One model. At list price Phi 4 falls in the llm11-fast band, but no pack routes to it because its context window is under 32K tokens, the floor for pack members. You can still pin it: pass microsoft/phi-4 as model and routing steps aside while verification still runs.
- What is a cheaper alternative to Phi 4?
- Nearest in price and below it: Qwen Qwen3 30B A3B Instruct 2507 at $0.048 in and $0.193 out, Google Gemma 3 12B at $0.050 in and $0.150 out and OpenAI gpt-oss-120b at $0.037 in and $0.170 out. The catalogue has no benchmark score for Phi 4, so this is a price ordering. It does not claim these answer as well.