Vercel AI Gateway 上的 Claude Opus 5

Try Claude Opus 5 via Vercel AI Gateway within ~$5 monthly free credits (when free-tier eligible). Rate-limited — not for production.

Updated 7/25/2026Provider trust 98/1003 code examples

About the Model

Why Claude Opus 5?

Anthropic's 2026-07-24 Opus upgrade: near Claude Fable 5 intelligence at half Fable's list price (same $5 / $25 per MTok as Opus 4.8). A strong default for long-horizon coding and knowledge work.

Strengths

  • Same sticker, higher unit output on official task-cost curves (Frontier-Bench, CursorBench, OSWorld, etc.).
  • Stronger self-check / iteration for multi-file and end-to-end agent tasks.
  • 1M context + 128k max output.
  • Optional Fast mode (~2.5× speed, 2× price).

Free basis (this directory)

Via Vercel AI Gateway: unpaid teams typically get ~$5 credits / 30 days usable on anthropic/claude-opus-5 when free-tier eligible. Enough for smoke tests, not unlimited agents. Strict rate limits — not for production.

Scope

Multilingual text + image input, tool use, agentic coding. Default on Claude Max; strongest model on Claude Pro.

Deep dive: MuiRouter.

How to Access for Free (via Vercel AI Gateway)

Free trial notes

Vercel AI Gateway grants unpaid teams roughly $5 credits every 30 days at provider list prices (no Gateway markup).

  • Model id: anthropic/claude-opus-5
  • Opus 5 list ~$5 / $25 per MTok — $5 is for short tests only
  • Credits usually start after the team's first AI Gateway request
  • Purchasing credits typically ends the free monthly grant — keep a Hobby team for trials
  • Free tier may be a subset of models; check Free Tier models
  • Strict rate limits: fine for day-to-day testing, not production

Setup: Vercel API key guide.

Try it in your browser

Pick a model, paste your own free key, and run. Your key is sent once to call the provider and never stored on our servers.

Pick a model, paste your own free key, and run. Your key is sent once to call the provider and never stored on our servers.

Code Examples

curl
curl https://ai-gateway.vercel.sh/v1/chat/completions \
  -H "Authorization: Bearer $AI_GATEWAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-opus-5",
    "messages": [{"role":"user","content":"Summarize this PR in 5 bullets."}]
  }'
python
from openai import OpenAI

client = OpenAI(
    api_key="...",  # Vercel AI Gateway key
    base_url="https://ai-gateway.vercel.sh/v1",
)
resp = client.chat.completions.create(
    model="anthropic/claude-opus-5",
    messages=[{"role": "user", "content": "Summarize this PR in 5 bullets."}],
)
print(resp.choices[0].message.content)
javascript
import { streamText } from 'ai'

const result = streamText({
  model: 'anthropic/claude-opus-5',
  prompt: 'Implement the multi-file refactor described in this issue.',
  reasoning: 'medium',
})
for await (const part of result.textStream) process.stdout.write(part)

Limitations & Caveats

Caveats

  • Output tokens are expensive; $5 burns quickly
  • Safety classifiers may block some security workloads — configure fallbacks
  • Buying credits usually ends the free monthly grant
  • Always trust the live Vercel dashboard over third-party summaries