> ## Documentation Index
> Fetch the complete documentation index at: https://inference-docs.cerebras.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Gemma 4 31B

> This model excels at multimodal reasoning across screenshots, documents, diagrams, and design assets. Ideal for visual agentic workflows, image-aware copilots, and teams migrating from closed multimodal APIs to an open model.

export const ModelInfo = ({modelId, modelCardUrl, playgroundUrl, contextLength = {}, maxOutput = {}, speed, inputOutput = {}, pricing = {}, rateLimits = [], endpoints = [], features = [], knownLimitations = []}) => {
  const [copied, setCopied] = React.useState(false);
  const hasImageLimits = rateLimits.some(limit => limit.imagesPerRequest);
  const rateLimitGridTemplate = hasImageLimits ? 'minmax(0, 1.2fr) repeat(4, minmax(0, 1fr))' : 'minmax(0, 1.2fr) repeat(3, minmax(0, 1fr))';
  const handleCopy = () => {
    navigator.clipboard.writeText(modelId);
    setCopied(true);
    setTimeout(() => setCopied(false), 2000);
  };
  return <div className="space-y-6 not-prose">
      {modelId && <div className="flex flex-col sm:flex-row sm:items-center sm:justify-between gap-3 border border-zinc-200 dark:border-zinc-800 rounded-xl px-5 py-3 bg-[#f7f5f2] dark:bg-zinc-900">
          <div className="flex items-center gap-3 min-w-0">
            <span className="text-xs font-mono font-semibold tracking-wider text-zinc-500 dark:text-zinc-400 uppercase shrink-0">Model ID</span>
            <button onClick={handleCopy} className="inline-flex items-center gap-1.5 group cursor-pointer bg-transparent border-0 p-0 min-w-0" title="Copy model ID">
              <code className="text-sm font-mono font-semibold text-zinc-900 dark:text-white break-all">{modelId}</code>
              <span className="text-zinc-400 dark:text-zinc-500 group-hover:text-zinc-600 dark:group-hover:text-zinc-300 transition-colors shrink-0">
                <Icon icon={copied ? "check" : "copy"} size={13} color="currentColor" />
              </span>
            </button>
          </div>
          <div className="flex items-center gap-4 sm:gap-6 shrink-0">
            {playgroundUrl && <a href={playgroundUrl} target="_blank" rel="noopener noreferrer" className="inline-flex items-center gap-1 text-sm text-zinc-700 dark:text-zinc-300 hover:text-zinc-900 dark:hover:text-white transition-colors">
                Try in Playground
                <Icon icon="arrow-right" size={14} color="currentColor" />
              </a>}
            {modelCardUrl && <a href={modelCardUrl} target="_blank" rel="noopener noreferrer" className="inline-flex items-center gap-1 text-sm text-zinc-700 dark:text-zinc-300 hover:text-zinc-900 dark:hover:text-white transition-colors">
                Model card
                <Icon icon="external-link" size={14} color="currentColor" />
              </a>}
          </div>
        </div>}

      <div className="bg-white dark:bg-zinc-950 border border-zinc-200 dark:border-zinc-800 rounded-xl p-6">
        <h3 className="text-base font-bold text-zinc-900 dark:text-white mb-6">Model Stats</h3>
        <div className="grid grid-cols-1 sm:grid-cols-2 md:grid-cols-4 gap-6">
          <div>
            <div className="text-xs font-mono font-semibold tracking-wider text-zinc-500 dark:text-zinc-400 mb-3 uppercase">SPEED</div>
            <div className="text-3xl font-bold text-orange-500 dark:text-orange-400">{speed?.value}</div>
            <div className="text-sm text-zinc-500 dark:text-zinc-400 mt-1">{speed?.unit}</div>
          </div>

          <div>
            <div className="text-xs font-mono font-semibold tracking-wider text-zinc-500 dark:text-zinc-400 mb-3 uppercase">CONTEXT WINDOW</div>
            <div className="space-y-2">
              <div className="flex items-baseline gap-3">
                <span className="text-xs text-zinc-500 dark:text-zinc-400 w-8 shrink-0">Free</span>
                <span className="text-sm font-semibold text-zinc-900 dark:text-white">{contextLength.freeTier}</span>
              </div>
              <div className="flex items-baseline gap-3">
                <span className="text-xs text-zinc-500 dark:text-zinc-400 w-8 shrink-0">Paid</span>
                <span className="text-sm font-semibold text-zinc-900 dark:text-white">{contextLength.paidTiers}</span>
              </div>
            </div>
          </div>

          <div>
            <div className="text-xs font-mono font-semibold tracking-wider text-zinc-500 dark:text-zinc-400 mb-3 uppercase">MAX OUTPUT</div>
            <div className="space-y-2">
              <div className="flex items-baseline gap-3">
                <span className="text-xs text-zinc-500 dark:text-zinc-400 w-8 shrink-0">Free</span>
                <span className="text-sm font-semibold text-zinc-900 dark:text-white">{maxOutput.freeTier || 'N/A'}</span>
              </div>
              <div className="flex items-baseline gap-3">
                <span className="text-xs text-zinc-500 dark:text-zinc-400 w-8 shrink-0">Paid</span>
                <span className="text-sm font-semibold text-zinc-900 dark:text-white">{maxOutput.paidTiers || 'N/A'}</span>
              </div>
            </div>
          </div>

          <div>
            <div className="text-xs font-mono font-semibold tracking-wider text-zinc-500 dark:text-zinc-400 mb-3 uppercase">MODALITY</div>
            <div className="space-y-2">
              <div className="flex items-baseline gap-3">
                <span className="text-xs text-zinc-500 dark:text-zinc-400 w-10 shrink-0">Input</span>
                <span className="text-sm font-semibold text-zinc-900 dark:text-white">
                  {inputOutput.inputFormats ? inputOutput.inputFormats.map(f => f.charAt(0).toUpperCase() + f.slice(1)).join(', ') : 'Text'}
                </span>
              </div>
              <div className="flex items-baseline gap-3">
                <span className="text-xs text-zinc-500 dark:text-zinc-400 w-10 shrink-0">Output</span>
                <span className="text-sm font-semibold text-zinc-900 dark:text-white">
                  {inputOutput.outputFormats ? inputOutput.outputFormats.map(f => f.charAt(0).toUpperCase() + f.slice(1)).join(', ') : 'Text'}
                </span>
              </div>
            </div>
          </div>
        </div>
      </div>

      {pricing.inputPrice && pricing.outputPrice && <div className="bg-white dark:bg-zinc-950 border border-zinc-200 dark:border-zinc-800 rounded-xl p-6">
          <div className="flex items-baseline gap-2 mb-6">
            <h3 className="text-base font-bold text-zinc-900 dark:text-white">Pricing</h3>
            <span className="text-sm text-zinc-500 dark:text-zinc-400">per million tokens</span>
          </div>
          <div className="flex flex-col sm:flex-row">
            <div className="flex-1">
              <div className="text-xs font-mono font-semibold tracking-wider text-zinc-500 dark:text-zinc-400 mb-2 uppercase">Input</div>
              <div className="text-4xl font-bold text-zinc-900 dark:text-white">{pricing.inputPrice.split(' /')[0]}</div>
            </div>
            <div className="hidden sm:block w-px bg-zinc-200 dark:bg-zinc-800 mx-8"></div>
            <div className="sm:hidden h-px bg-zinc-200 dark:bg-zinc-800 my-5"></div>
            <div className="flex-1">
              <div className="text-xs font-mono font-semibold tracking-wider text-zinc-500 dark:text-zinc-400 mb-2 uppercase">Output</div>
              <div className="text-4xl font-bold text-zinc-900 dark:text-white">{pricing.outputPrice.split(' /')[0]}</div>
            </div>
          </div>
          <p className="text-sm text-zinc-500 dark:text-zinc-400 mt-10">
            Developer pricing. For volume discounts and enterprise features, see our{' '}
            <a href="https://www.cerebras.ai/pricing" className="text-black dark:text-white font-semibold underline decoration-orange-500 underline-offset-4 decoration-1 hover:decoration-2">
              pricing page
            </a>
            .
          </p>
        </div>}

      {knownLimitations && knownLimitations.length > 0 && <div className="bg-white dark:bg-zinc-950 border border-zinc-200 dark:border-zinc-800 rounded-xl p-6">
          <h3 className="text-base font-bold text-zinc-900 dark:text-white mb-4">Model Notes</h3>
          <div className="space-y-2">
            {knownLimitations.map((limitation, index) => <div key={index} className="flex items-start gap-2 py-2">
                <div className="w-1.5 h-1.5 bg-zinc-400 dark:bg-zinc-500 rounded-full flex-shrink-0 mt-2"></div>
                <div className="text-zinc-900 dark:text-white text-sm leading-relaxed prose-sm max-w-none">
                  {limitation}
                </div>
              </div>)}
          </div>
        </div>}

      {rateLimits && rateLimits.length > 0 && <div className="bg-white dark:bg-zinc-950 border border-zinc-200 dark:border-zinc-800 rounded-xl p-6">
          <h3 className="text-base font-bold text-zinc-900 dark:text-white mb-4">Rate Limits</h3>
          <div role="table" aria-label="Rate limits" className="w-full text-sm">
            <div role="row" className="grid border-b border-zinc-200 dark:border-zinc-800" style={{
    gridTemplateColumns: rateLimitGridTemplate
  }}>
              <div role="columnheader" className="text-left py-3 px-2 text-zinc-500 dark:text-zinc-400 font-medium">Tier</div>
              <div role="columnheader" className="whitespace-normal break-words text-right leading-tight py-3 px-2 text-zinc-500 dark:text-zinc-400 font-medium">Requests / min</div>
              <div role="columnheader" className="whitespace-normal break-words text-right leading-tight py-3 px-2 text-zinc-500 dark:text-zinc-400 font-medium">Input tokens / min</div>
              <div role="columnheader" className="whitespace-normal break-words text-right leading-tight py-3 px-2 text-zinc-500 dark:text-zinc-400 font-medium">Daily tokens</div>
              {hasImageLimits && <div role="columnheader" className="whitespace-normal break-words text-right leading-tight py-3 px-2 text-zinc-500 dark:text-zinc-400 font-medium">Images / request</div>}
            </div>
            <div role="rowgroup">
              {rateLimits.map((limit, index) => <div key={index} role="row" className="grid border-b border-zinc-200/50 dark:border-zinc-800/50 last:border-b-0" style={{
    gridTemplateColumns: rateLimitGridTemplate
  }}>
                  <div role="cell" className="py-3 px-2 text-zinc-900 dark:text-white font-medium">{limit.tier}</div>
                  <div role="cell" className="text-right py-3 px-2 text-zinc-900 dark:text-white">{limit.requestsPerMin}</div>
                  <div role="cell" className="text-right py-3 px-2 text-zinc-900 dark:text-white">{limit.inputTokensPerMin}</div>
                  <div role="cell" className="text-right py-3 px-2 text-zinc-900 dark:text-white">{limit.dailyTokens}</div>
                  {hasImageLimits && <div role="cell" className="text-right py-3 px-2 text-zinc-900 dark:text-white">{limit.imagesPerRequest || 'N/A'}</div>}
                </div>)}
            </div>
          </div>
        </div>}

      {(endpoints && endpoints.length > 0 || features.length > 0) && <div className="grid grid-cols-1 sm:grid-cols-2 gap-6">
          {endpoints && endpoints.length > 0 && <div className="bg-white dark:bg-zinc-950 border border-zinc-200 dark:border-zinc-800 rounded-xl p-6">
              <h3 className="text-base font-bold text-zinc-900 dark:text-white mb-4">Endpoints</h3>
              <ul className="space-y-2">
                {endpoints.map((endpoint, index) => {
    const endpointUrls = {
      'Chat Completions': '/v1/chat/completions',
      'Completions': '/v1/completions',
      'Models': '/v1/models'
    };
    const endpointName = typeof endpoint === 'string' ? endpoint : endpoint.name;
    const endpointUrl = typeof endpoint === 'object' && endpoint.url ? endpoint.url : endpointUrls[endpointName];
    return <li key={index} className="flex items-center gap-2 text-sm text-zinc-900 dark:text-white">
                      <div className="w-1.5 h-1.5 bg-zinc-400 dark:bg-zinc-500 rounded-full flex-shrink-0"></div>
                      <span className="font-medium">{endpointName}</span>
                      {endpointUrl && <code className="text-xs text-zinc-500 dark:text-zinc-400 font-mono">{endpointUrl}</code>}
                    </li>;
  })}
              </ul>
            </div>}
          {features.length > 0 && <div className="bg-white dark:bg-zinc-950 border border-zinc-200 dark:border-zinc-800 rounded-xl p-6">
              <h3 className="text-base font-bold text-zinc-900 dark:text-white mb-4">Capabilities</h3>
              <ul className="space-y-2">
                {features.map((feature, index) => <li key={index} className="flex items-center gap-2 text-sm text-zinc-900 dark:text-white">
                    <div className="w-1.5 h-1.5 bg-zinc-400 dark:bg-zinc-500 rounded-full flex-shrink-0"></div>
                    {feature}
                  </li>)}
              </ul>
            </div>}
        </div>}


      <div className="bg-gradient-to-r from-orange-600/10 to-red-500/10 border border-orange-600/20 rounded-xl p-6">
        <div className="flex items-center gap-3 mb-3">
          <div className="w-8 h-8 bg-orange-600/20 rounded-lg flex items-center justify-center">
            <Icon icon="rocket" size={18} color="#fb923c" />
          </div>
          <h3 className="text-lg font-semibold text-zinc-900 dark:text-white">Need Higher Limits?</h3>
        </div>
        <p className="text-zinc-700 dark:text-zinc-300">
          Reach out for custom pricing with our Enterprise tier for higher rate limits and dedicated support.
        </p>
        <div className="mt-4">
          <a href="https://cerebras.ai/contact-us" className="inline-flex items-center gap-2 bg-orange-600 hover:bg-orange-700 text-white px-4 py-2 rounded-lg transition-colors font-medium">
            Contact Sales
            <Icon icon="arrow-right" color="white" size={16} />
          </a>
        </div>
      </div>
    </div>;
};

Model ID: `gemma-4-31b`. Speed: \~1500 tokens/sec. Context window: 65k tokens (free tier), 131k tokens (paid). Max output: 32k tokens (free tier), 40k tokens (paid). Pricing: $2.15 per million input tokens, $2.70 per million output tokens. Modality: multimodal. Accepts text and image inputs (base64 PNG or JPEG data URI only; external URLs not supported), produces text output. Max images per request: 2 for Free Trial, 10 for Developer and Enterprise accounts on the shared tier. Higher image limits may be available with Dedicated Endpoints or explicit organization configurations. Max 10 MB total image payload. Image inputs are only available in Chat Completions; the Completions endpoint does not support images. Capabilities: Image Inputs, Reasoning, Streaming, Sampling Controls, Structured Outputs, Tool Calling, Parallel Tool Calling, Prompt Caching. Reasoning is disabled by default; enable it with the `reasoning_effort` parameter. The `raw` and `hidden` reasoning formats are not supported. Structured outputs and tool calling with `strict: true` (constrained decoding) are supported. Rate limits (free tier): 5 requests/min, 30k input tokens/min, 1M tokens/day. Rate limits (Pay as You Go): 300 requests/min, 500k input tokens/min. Notes: Recommended starting parameters: temperature=1.0, top\_p=0.95.

<ModelInfo
  modelId="gemma-4-31b"
  modelCardUrl="https://huggingface.co/google/gemma-4-31B-it"
  playgroundUrl="https://cloud.cerebras.ai"
  contextLength={{
freeTier: "65k tokens",
paidTiers: "131k tokens"
}}
  maxOutput={{
freeTier: "32k tokens",
paidTiers: "40k tokens"
}}
  speed={{
value: "~1850",
unit: "tokens/sec"
}}
  pricing={{
inputPrice: "$0.99",
outputPrice: "$1.49"
}}
  rateLimits={[
{
  tier: "Free Trial",
  requestsPerMin: "5",
  inputTokensPerMin: "30k",
  dailyTokens: "1M",
  imagesPerRequest: "2"
},
{
  tier: "Developer",
  requestsPerMin: "300",
  inputTokensPerMin: "500k",
  dailyTokens: "N/A",
  imagesPerRequest: "10"
}
]}
  endpoints={[
"Chat Completions",
"Completions"
]}
  features={[
"Image Inputs",
"Reasoning",
"Streaming",
"Sampling Controls",
"Structured Outputs",
"Tool Calling",
"Parallel Tool Calling",
"Prompt Caching"
]}
  inputOutput={{
inputFormats: ["text", "image"],
outputFormats: ["text"]
}}
  knownLimitations={[
 <span>
  Image inputs are only available in Chat Completions. The Completions endpoint does not support image inputs.
</span>,
<span>
  See the <a href="/capabilities/image-inputs" className="font-semibold text-zinc-900 dark:text-white underline underline-offset-4 decoration-2 hover:text-orange-500 hover:decoration-4" style={{ textDecorationColor: '#f97316' }}>Image Inputs guide</a> for limits and token behavior.
</span>,
<span>
Structured outputs and tool calling with <code>strict: true</code> (constrained decoding) is supported for this model.
</span>,
<span>
  Reasoning is disabled by default for Gemma 4. Use <code>reasoning_effort</code> to control it in Chat Completions. Gemma 4 does not support <code>raw</code> or <code>hidden</code> reasoning formats today. See the <a href="/capabilities/reasoning#gemma-4:-reasoning_effort" className="font-semibold text-zinc-900 dark:text-white underline underline-offset-4 decoration-2 hover:text-orange-500 hover:decoration-4" style={{ textDecorationColor: '#f97316' }}>reasoning guide</a> for details.
</span>
]}
/>