Phi-2 on Fireworks AI

Phi-2 · Microsoft Research

ProvisionedOpen Source

Last refreshed 2026-05-19. Next refresh: weekly.

Why use Phi-2 on Fireworks AI?

Fireworks AI offers Phi-2 with pay-as-you-go pricing at $0.10/1M input tokens. Fireworks AI offers a generative AI platform as a service, focusing on rapid product iteration and cost-efficient AI deployment.

Compare Phi-2 across 5 providers to find the best fit for your use case
Input / 1M
$0.10
Output / 1M
$0.10
Cache
Not sourced
Batch
Not sourced

Setup recipe

Python + curl
Install
pip install openai
Auth
export FIREWORKS_API_KEY=...
Call
import os
from openai import OpenAI
client = OpenAI(
    api_key=os.environ["FIREWORKS_API_KEY"],
Model ID
phi-2

Request example

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["FIREWORKS_API_KEY"],
    base_url="https://api.fireworks.ai/inference/v1"
)
response = client.chat.completions.create(
    model="phi-2",
    messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)

Gotchas

  • Fireworks model IDs use "accounts/fireworks/models/{model-name}" format, e.g. "accounts/fireworks/models/llama4-scout-instruct-basic" or "accounts/fireworks/models/deepseek-r1".
  • The examples expect FIREWORKS_API_KEY; rename it only if your application config maps the new variable.

Compare Phi-2 Across Providers

ProviderInput (per 1M)Output (per 1M)
Microsoft Foundry$0.07$0.07
Cloudflare Workers AI——
Together AI$0.10$0.10
Fireworks AI$0.10$0.10
Replicate API$0.05$0.25

Pricing

TypePrice (per 1M)
Input tokens$0.10
Output tokens$0.10

Capabilities

Structured Outputs

About Phi-2

Phi-2 is a compact language model by Microsoft endowed with 2.7 billion parameters and part of their Phi series. It shows formidable capabilities in reasoning and language understanding, outshining much larger models, even those with up to 25 times more parameters. Phi-2's training utilized a vast and diverse dataset of 1.4 trillion tokens, incorporating high-quality synthetic data and curated web content to bolster its common sense reasoning and general knowledge. Interestingly, despite lacking fine-tuning via reinforcement learning from human feedback (RLHF), it exhibits enhanced safety features and reduced bias.

Get Started