Llama 4 Maverick API pricing

Meta's Llama 4 Maverick. Available at listed at platform rate (no published official USD price).

USD, standard tier, verified 2026-09-26.

Specifications

What Llama 4 Maverick costs at your volume

Calculated at $0.07 input / $0.35 output per 1M tokens, assuming a 5:1 input-to-output ratio. Your ratio decides the real number — run your own figures through the cost calculator.

When Llama 4 Maverick is the right call

At $0.07 per million input tokens this sits in the low-cost band. It is a good fit for high-volume work — classification, extraction, summarisation at scale, autocomplete, and any pipeline where you call the model thousands of times a day.

It is usually the wrong call for tasks where a wrong answer is expensive to unwind; at this price the saving is small enough that escalating a hard case to a stronger model is usually worth it.

Where Llama 4 Maverick sits in the index

One model, three rankings — because the model that is cheapest per input token is frequently not the cheapest per task. Output tokens cost a median of 4.0× what input costs across these 216 models, so a workload that generates more than it reads flips the order. Percentile is the share of models that cost more than this one.

On an agent mix the lowest blended rate in the same index is Gemma2 9B It at $0.01 per 1M tokens — about 28.0× less than this model at the same ratio. That gap is what the re-ranked index exists to show: sorting by input price alone can put a model 68 places out.

Similarly priced alternatives

Blended comparison (input + output per 1M). Price is one axis — the models above are not interchangeable, they are the ones worth testing side by side at this budget.

Calling Llama 4 Maverick

The identifier you send is llama-4-maverick — copy it exactly, because a mismatch returns a model error rather than a connection error, which sends most people to debug the wrong thing. Everything else is a base URL argument; official SDKs work unchanged.

from openai import OpenAI

client = OpenAI(base_url="https://aicomp.ai/v1", api_key="sk-...")

resp = client.chat.completions.create(
    model="llama-4-maverick",
    messages=[
        {"role": "system", "content": "Answer in one short paragraph."},
        {"role": "user", "content": "Explain what changed in this release."},
    ],
    max_tokens=300,
)

# prompt_tokens / completion_tokens 是真实账单的唯一依据 —— 记下来，别猜
print(resp.usage.prompt_tokens, resp.usage.completion_tokens)

Log the usage object on every call, not just while you are optimising. Rates follow vendor promotions and can move without notice, so a per-call record is the only thing that tells you later whether a price change actually hit you — and in which direction.

import json

# 把每次调用的 usage 落盘，一周后你就有了自己的 in:out 比例。
# 拿这个比例去 /calculator/ 比价，比对着价目表猜准得多 ——
# output 单价是 input 的好几倍，猜错方向就是往贵了算。
with open("usage.jsonl", "a") as fh:
    fh.write(json.dumps({
        "model": model,
        "in": resp.usage.prompt_tokens,
        "out": resp.usage.completion_tokens,
    }) + "\n")

Setting it up

Pointing an existing integration at this model is a base URL change, not a rewrite. Pick your tool:

• Claude Code — set the base URL via environment variables
• Cursor — override the API endpoint in settings
• OpenAI / Anthropic SDK — one base_url argument
• Python — sync, streaming and async examples
• Cutting the bill — eight levers ranked by effect

FAQ

Get Llama 4 Maverick access