Baltor Get started

Models

Llama-4-Scout-17B-16E-Instruct

By meta-llama. 108.6 billion parameters, a context window of 10,485,760 tokens and the licence other.

Facts

Released
2025-04-02 the day the repository was first published on Hugging Face Hugging Face, read
Licence
other the licence the model card declares Hugging Face, read
Open weights
Yes the weights are published in this Hugging Face repository, behind the maker's access form Hugging Face, read
Parameters
108.6 billion counted from the safetensors weight files Hugging Face, read
Active parameters
18.0 billion estimate estimated from the expert counts and sizes in the configuration model configuration, read
Knowledge cutoff
Unknown
Context window
  • 10,485,760 tokens the maximum position embeddings in the model configuration model configuration, read
Longest output
Unknown
Tool calling
Unknown
Structured output
Unknown
Reasoning controls
Unknown
Inputs and outputs
text, image in, text out Hugging Face, read
Good for
  • Vision: Hugging Face files it under image-text-to-text Hugging Face, read

Prices

Each provider's own price in US dollars per million tokens, as an openly licensed source lists it
ProviderInputOutputContextAs ofSource
nscale0.090 USD0.290 USDUnknownLiteLLM, read
Deep Infra0.100 USD0.300 USD327,680 older than 30 days; check the providermodels.dev, read
Novita AI0.180 USD0.590 USD131,072 older than 30 days; check the providermodels.dev, read
Together AI0.180 USD0.590 USD1,048,576LiteLLM, read

Run it on your own hardware

Weights are the sizes of the files a source lists, or an estimate from the parameter count where none does. Memory is an estimate: weights plus KV cache plus 512 MiB and 5 percent of the weights for runtime buffers. Check it against your hardware.

QuantizationWeightsMemory at 8,192 tokensMemory at 32,768 tokens
IQ1_M24.5 GiB27.7 GiB32.2 GiB
IQ2_XXS28.1 GiB31.5 GiB36.0 GiB
IQ2_XS30.7 GiB34.2 GiB38.7 GiB
IQ2_S32.0 GiB35.6 GiB40.1 GiB
IQ2_M34.6 GiB38.3 GiB42.8 GiB
Q2_K40.0 GiB44.0 GiB48.5 GiB
Q2_K_L41.0 GiB45.0 GiB49.5 GiB
IQ3_XXS41.9 GiB46.0 GiB50.5 GiB
IQ3_XS44.2 GiB48.4 GiB52.9 GiB
Q3_K_S46.3 GiB50.7 GiB55.2 GiB
IQ3_M46.9 GiB51.2 GiB55.7 GiB
Q3_K_M50.6 GiB55.1 GiB59.6 GiB
Q3_K_L53.8 GiB58.5 GiB63.0 GiB
Q3_K_XL54.7 GiB59.4 GiB63.9 GiB
IQ4_XS55.8 GiB60.6 GiB65.1 GiB
IQ4_NL58.7 GiB63.6 GiB68.1 GiB
Q4_058.7 GiB63.7 GiB68.2 GiB
Q4_K_M62.9 GiB68.1 GiB72.6 GiB
Q4_K_L63.6 GiB68.8 GiB73.3 GiB
Q4_164.4 GiB69.6 GiB74.1 GiB
Q5_K_L73.9 GiB79.6 GiB84.1 GiB
Q6_K_L83.1 GiB89.3 GiB93.8 GiB
Q8_0106 GiB113 GiB117 GiB
BF16201 GiB213 GiB217 GiB

Start it with a local runtime

  • Ollama
    ollama run hf.co/bartowski/meta-llama_Llama-4-Scout-17B-16E-Instruct-old-GGUF:Q4_0
  • LM Studio
    lms get bartowski/meta-llama_Llama-4-Scout-17B-16E-Instruct-old-GGUF
    lms server start
  • llama.cpp server
    llama-server -hf bartowski/meta-llama_Llama-4-Scout-17B-16E-Instruct-old-GGUF:Q4_0 --jinja
  • vLLM
    vllm serve meta-llama/Llama-4-Scout-17B-16E-Instruct
  • SGLang
    sglang serve --model-path meta-llama/Llama-4-Scout-17B-16E-Instruct --port 30000

Use it from your harness

Set up for DeepInfra with the model meta-llama/Llama-4-Scout-17B-16E-Instruct. Each endpoint page has the same setup for its own address.

OpenCode

Put this in opencode.json in your project folder:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "deepinfra": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "DeepInfra",
      "options": {
        "baseURL": "https://api.deepinfra.com/v1/openai",
        "apiKey": "{env:DEEPINFRA_TOKEN}"
      },
      "models": {
        "meta-llama/Llama-4-Scout-17B-16E-Instruct": {
          "name": "Llama-4-Scout-17B-16E-Instruct"
        }
      }
    }
  }
}
  • OpenCode reads any OpenAI-compatible address through the @ai-sdk/openai-compatible package, and an address that speaks the Responses API through @ai-sdk/openai.

From OpenCode documentation, read .

Pi

Put this in ~/.pi/agent/models.json:

{
  "providers": {
    "deepinfra": {
      "baseUrl": "https://api.deepinfra.com/v1/openai",
      "api": "openai-completions",
      "apiKey": "$DEEPINFRA_TOKEN",
      "models": [
        {
          "id": "meta-llama/Llama-4-Scout-17B-16E-Instruct"
        }
      ]
    }
  }
}
  • The apiKey field can name an environment variable as $NAME.

From Pi documentation, read .

Codex

Codex speaks only the Responses API, and DeepInfra documents no Responses address. A gateway that offers one can sit in between.

  • Codex speaks the Responses API only: responses is the one supported wire API of a custom provider. Ollama and LM Studio are built in and start with --oss.

From Codex documentation, read .

Claude Code

Put this in ~/.claude/settings.json, or variables in your shell:

export ANTHROPIC_BASE_URL="https://api.deepinfra.com/anthropic"
export ANTHROPIC_AUTH_TOKEN="$DEEPINFRA_TOKEN"
export ANTHROPIC_MODEL="meta-llama/Llama-4-Scout-17B-16E-Instruct"
claude
  • Claude Code sends Anthropic Messages requests to ANTHROPIC_BASE_URL. Anthropic says it does not support routing Claude Code to models other than Claude through any gateway, so some features may not work with another model.

From Claude Code documentation, read .

Other ways to reach it

  • Novita AI, model meta-llama/llama-4-scout-17b-16e-instruct
  • Together AI, model meta-llama/Llama-4-Scout-17B-16E-Instruct
  • Ollama, model hf.co/bartowski/meta-llama_Llama-4-Scout-17B-16E-Instruct-old-GGUF:Q4_0

Published results

These are results other people published. Baltor did not run them and does not rank models by them.

Two more publishers are linked, never copied, because their terms do not allow it:

  • Artificial Analysis publishes its own intelligence, speed and price measurements.
  • OpenRouter lists current prices and uptime for each provider it routes to.