Meta AI makes Llama 4 Maverick 17B 128E Instruct, an open-source multimodal chat model that takes in text and images and outputs text. It is a mixture-of-experts model designed for tasks like generating code, calling functions, and producing structured outputs.
Input
Output
Context
1M
Max Output
-
Parameters
401.6B
Input Modalities
Output Modalities
Some values come from web research and may be incorrect. Report incorrect data
Measured speed: 48 tok/s output speed · 0.58 s to first answer token · 0.58 s to first token
Bands compare this model with every model the same suite measured in our catalog.
Download 417 GB · Runs on vLLM v0.32.0
| Precision | Weights | One request at 33K tokens | Smallest GPU that fits |
|---|---|---|---|
| FP16 / BF16 | 748 GB | 800 GB | 4x AMD Instinct MI325X (1024 GB) |
| INT8 | 374 GB | 407 GB | 2x AMD Instinct MI325X (512 GB) |
| INT4 | 187 GB | 210 GB | 1x AMD Instinct MI325X (256 GB) |
Computed by Inferbase from the 401,583,781,376 parameters in the creator's Hugging Face repository; one request adds its KV cache, activations and framework overhead, with the layer layout estimated from the parameter count.
Request meta-llama-4-maverick-17b-128e-instruct and routing picks a build. Use a build's model ID to pin it.
| Build | What it is | Model ID |
|---|---|---|
| FP8 | 8-bit floating-point weights | meta-llama-4-maverick-17b-128e-instruct:fp8 |
Public resources
Specification sources
As of Nov 14, 2026.
Answered from the facts on this page, with their sources.
Llama 4 Maverick 17B 128E Instruct has a context window of 1,048,576 tokens (source: Creator or provider API). A provider may serve less; Inferbase routes on what each provider serves.
Yes. Llama 4 Maverick 17B 128E Instruct is an open-weight model under the Other license. You can download and self-host it within the license terms.
Llama 4 Maverick 17B 128E Instruct has 401,583,781,376 parameters (401.6B), as stated in the creator's Hugging Face repository.
Strengths: 1,048K token context window; Supports 12 languages. Limitations: Below average on graduate-level science (GPQA) among measured models; Below average on competition math (AIME 2025) among measured models; Below average on long-context reasoning among measured models. Each is computed from the model's stated facts and benchmark results.
Llama 4 Maverick 17B 128E Instruct accepts Text and Image and produces Text. The Capabilities card lists the API features it supports, such as function calling and structured output.
Llama 4 Maverick 17B 128E Instruct has published results from Artificial Analysis and Epoch AI. The Benchmarks card shows each score as reported, with its band among the models the same source measured. Scores are never combined across sources.
Yes. Llama 4 Maverick 17B 128E Instruct is available on Inferbase Managed through the OpenAI-compatible API as meta-llama-4-maverick-17b-128e-instruct. Request it by name, or request auto and routing picks it when it fits the request best.
The weights take 748.0 GB at FP16 / BF16, 374.0 GB at INT8, 187.0 GB at INT4, computed from the 401,583,781,376 parameters its creator publishes. Each request adds its KV cache and the framework's overhead. The hardware section shows one request and the smallest GPU that fits; the GPU planner sizes it for your context length and traffic.
The model card of Llama 4 Maverick 17B 128E Instruct lists 12 languages: Arabic, German, English, Spanish, French, Hindi, Indonesian, Italian, Portuguese, Thai, Filipino and Vietnamese. That is what Meta AI states; the catalog does not test language quality.
Start on the free plan with 5,000 requests a month, no card required.