Claude Opus 4.7 Is Now Live on 1vy

2026-04-18

Anthropic's newest flagship — Claude Opus 4.7 — is available through the 1vy gateway today. $5.75/$28.75 per 1M tokens, 1M context window at standard pricing, OpenAI SDK compatible. Plus the one tokenizer change you should know about before switching.

Claude Opus 4.7 — Anthropic's newest flagship model — is live on 1vy today. You can call it through the same OpenAI-compatible endpoint you already use for Sonnet 4.6, Haiku 4.5, GPT-5, Gemini 3 Pro, and 50+ other models. One API key, no separate Anthropic account, no SDK migration.

TL;DR

- Model ID: claude-opus-4-7 - Pricing on 1vy: $5.75 / 1M input tokens · $28.75 / 1M output tokens (15% markup over Anthropic's $5/$25 supplier price) - Context window: 1,000,000 tokens at standard pricing (no >200k premium) - One-line switch: change your model parameter from "claude-opus-4-6" to "claude-opus-4-7" and you're done - Important caveat: the new tokenizer can use up to 35% more tokens for the same text — your per-call cost may rise even though per-token pricing matches 4.6

How to call it

Already using the OpenAI SDK against 1vy? Just change one parameter. If you're new to 1vy, set base_url to https://api.1vy.ai/v1 and use a 1vy API key (sign up at 1vy.ai/register, $1 free credit on the house).

Python

```python import openai

client = openai.OpenAI( base_url="https://api.1vy.ai/v1", api_key="sk-1vy-your-key-here", )

response = client.chat.completions.create( model="claude-opus-4-7", messages=[ {"role": "system", "content": "You are a careful, thorough analyst."}, {"role": "user", "content": "Walk me through the trade-offs of Postgres vs SQLite for a multi-tenant SaaS at 1k tenants, 100M rows."}, ], ) print(response.choices[0].message.content) ```

Node.js

```javascript import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://api.1vy.ai/v1", apiKey: "sk-1vy-your-key-here", });

const resp = await client.chat.completions.create({ model: "claude-opus-4-7", messages: [ { role: "system", content: "You are a careful, thorough analyst." }, { role: "user", content: "What's the right way to scope a Postgres schema for a multi-tenant SaaS?" }, ], }); console.log(resp.choices[0].message.content); ```

cURL

``bash curl https://api.1vy.ai/v1/chat/completions \ -H "Authorization: Bearer sk-1vy-your-key-here" \ -H "Content-Type: application/json" \ -d '{ "model": "claude-opus-4-7", "messages": [ {"role": "user", "content": "Hello, Opus."} ] }' ``

Streaming, function calling, tool use, vision, and long-context all work the same as Opus 4.6 — 1vy proxies the SSE stream verbatim.

Pricing (and what's actually new under the hood)

1vy adds a flat 15% markup over Anthropic's published list price. No subscriptions, no minimums, no tiered overages.

This pricing is identical to what we already charge for Opus 4.6 and Opus 4.5 — Anthropic kept the Opus 4.x flagship tier at the same headline rate. (For reference, that's the same supplier-cost level we corrected in the April 2026 pricing audit — Opus 4.6 was previously mispriced at the old Opus 4 / 4.1 tier of $15/$75.)

Cost vs. Anthropic direct

Anthropic charges the same $5/$25 directly. Going through 1vy at $5.75/$28.75 costs about $0.75 more per million input tokens and $3.75 more per million output. For a typical 4k-input/2k-output chat completion, that's an extra $0.0105 per call. In return:

- One key, all providers: switch from Opus 4.7 to GPT-5 to DeepSeek R1 by changing the model parameter. No second account, no second invoice, no second rate limit to manage. - OpenAI SDK drop-in: skip the Anthropic Python/JS SDK entirely. Use whichever language you already have OpenAI bindings for. - Multi-provider failover for streaming chat (when a single upstream goes flaky, 1vy reroutes server-side). - Single billing: one invoice, one balance, one usage dashboard.

If you're calling Claude in single-vendor production at high volume, going direct can make sense. If you're A/B testing models, building a multi-model app, or just want the operational simplicity, 1vy pays for itself in saved engineering hours.

The tokenizer caveat (read this if you're cost-conscious)

Anthropic's Opus 4.7 release notes:

> Opus 4.7 uses a new tokenizer compared to previous models, contributing to its improved performance on a wide range of tasks. This new tokenizer may use up to 35% more tokens for the same fixed text.

What this means for you:

- Per-token pricing is the same as Opus 4.6 ($5.75/$28.75 on 1vy) - Per-call cost can be ~10-35% higher for the same English prompt, because the new tokenizer breaks text into more tokens - The trade-off Anthropic made: more tokens per word → finer-grained representation → measurably better quality on complex tasks

If you're switching from 4.6 to 4.7 and care about cost predictability, measure first. Send a representative prompt to both, compare usage.input_tokens in the response. For most prompts the difference will be in the single-digit-percent range; some text (especially code or non-English content) sees more.

Pricing comparison: Opus 4.7 vs the alternatives

For a one-million-input-token, 250k-output-token workload (a typical heavy reasoning session):

(Calculated as input_M × $/M + 0.25 × output_M × $/M for 1M input / 250k output.)

Opus 4.7 sits at the premium quality tier. If you're handling ambiguous reasoning, complex coding, or long-context analysis where output quality dominates the cost calculation, the per-token premium is justified. For high-volume production traffic where good-enough quality is fine, Sonnet 4.6, Gemini 3 Pro, or DeepSeek R1 all deliver more value per dollar.

See the full model catalog for live pricing across all 55 models.

What's new in 4.7 (vs 4.6)

Anthropic hasn't published a detailed changelog yet, but the things observable from the public docs:

1. New tokenizer (covered above) — primary architectural change driving the quality bump 2. Same 1M context window at standard rates as 4.6 (no premium for >200k tokens — still rare in the industry) 3. Same fast-mode availability (research preview): 6× standard rate for premium-speed inference, applicable across the full context window 4. Same data residency options: inference_geo: us for US-only routing at +10% cost (relevant for HIPAA / FedRAMP workloads)

If you weren't using prompt caching or fast mode on 4.6, this is essentially a quality upgrade at the same per-token price.

FAQ

Is claude-opus-4-7 the right model_id to use on the 1vy API?

Yes — exactly that string. Don't add a date suffix; 1vy normalizes the model name and routes to the channel's mapped Anthropic API ID. If your channel admin uses model aliases, the underlying call may go to claude-opus-4-7-20260418 or similar, but you should always pass "claude-opus-4-7" from your code.

Will my existing Opus 4.6 code keep working?

Yes — Opus 4.6 is still active at the same price ($5.75/$28.75 on 1vy), with a tentative retirement date no sooner than February 5, 2027. Anthropic provides at least 60 days notice before any model retirement and posts public retirement dates per model in their deprecation table. If you don't need the 4.7 quality bump, no action required.

Does prompt caching work?

Yes. 1vy proxies Anthropic's cache_control directive verbatim. Cache writes are billed at 1.25× input price (5-min cache) or 2× input price (1-hour cache); cache reads at 0.1× input price. The same multipliers Anthropic publishes — 1vy's 15% markup applies on top.

Can I use it with the Anthropic SDK directly?

The 1vy endpoint speaks the OpenAI dialect. To use the Anthropic-native SDK, you'd want to call Anthropic directly (and lose the multi-provider routing benefits). Most teams find it easier to standardize on the OpenAI SDK and route through 1vy.

How do I get started?

Create a free 1vy account (no credit card, $1 free credit). Generate an API key in the dashboard. Set base_url to https://api.1vy.ai/v1 and model to "claude-opus-4-7". That's it.

Try it now

Create a free account → ($1 credit, no card required)

See full pricing for all 55 models →

Compare Opus 4.7 vs GPT-5 → ｜ Opus 4.7 vs Opus 4.6 →

---

Sources: pricing verified against Anthropic's official pricing page on 2026-04-18. The new-tokenizer note is a direct quote from the Anthropic Opus 4.7 release notes. For pricing changes to other models in this update, see the April 2026 pricing audit.