Every Claude API discount, side by side
| Discount | What it saves | The catch |
|---|---|---|
| Batch API | 50% on input and output | Results within 24 hours, so not for chat or agents |
| Prompt caching | Repeated input at 0.025x to 0.1x the input price | Writing to the cache costs 1.25x (5 minutes) or 2x (1 hour) |
| A smaller model | Haiku costs a fraction of Opus | Weaker on hard problems |
| Volume discount | Negotiated | For high-volume users, through Anthropic's sales |
| Startup credits | Up to $100K in API credits | Through a partner VC; the program isn't taking applications now |
| Plan API credits | $100 a month on Max, $200 on Ultra, up to $500 on Team | Expire each billing cycle and don't cover Claude Code |
| A lower price per token | 46–65% below list on Zurelay | No Batches or Files API; prepaid credit |
Batch API: half price if you can wait
The Batch API takes a file of requests and bills both input and output at half the list price. Batches that aren't done within 24 hours expire, though most finish within an hour. It suits evaluations, classification, summaries and data enrichment: anything nobody is waiting on. Caching discounts stack on top of it.
Prompt caching: the discount you may already get
A cache read costs 0.1x the input price on most Claude models, 0.05x on Opus 5.5 and Sonnet 5.5, and 0.025x on Fable 5.1 and 5.2. Writing to the cache costs 1.25x the input price for five minutes or 2x for an hour, so caching pays off once the same start of a prompt is read twice. Claude Code marks its prompts for caching on its own; in your own code, add cache_control. More in prompt caching explained.
Pick the cheapest model that does the job
The model is the biggest lever. Per output token, Claude Opus 5.5 costs 40 times as much as Claude Haiku 5.5 at Anthropic's list price. List prices per 1M tokens, with Zurelay's beside them:
| Model | Input | Output | On Zurelay (in / out) |
|---|---|---|---|
| Claude Fable 5.2 | $8.00 | $40.00 | $3.20 / $16.00 |
| Claude Opus 5.5 | $4.00 | $20.00 | $1.40 / $7.00 |
| Claude Sonnet 5.5 | $2.00 | $10.00 | $0.79 / $3.95 |
| Claude Haiku 5.5 | $0.10 | $0.50 | $0.04 / $0.20 |
Volume, startup and research deals
- Volume: Anthropic says volume discounts may be available to high-volume users, negotiated case by case through its sales team.
- Startups: founders backed by a VC in Anthropic's partner network may get up to $100K in API credits, claimed through the VC. The Claude for Startups program's current cohort is full, and it isn't taking new applications.
- Academic and research: Anthropic says discounts may be available; ask its sales team.
- Plan credits: Max includes $100 of API credit a month and Ultra $200. Pro includes none. More in how to get a Claude API key.
Is there a Claude API coupon code?
No. Anthropic's pricing pages list no coupon or promo codes for the API, so codes on coupon sites aren't Anthropic's. Be careful with offers of free or shared Claude keys: they get revoked, and some cheap endpoints answer with a different model under Claude's name (how to check).
Pay less for every token
Zurelay sells every Claude model below Anthropic's list price (46–65% below), through the same Messages API and an OpenAI-compatible endpoint, from $10 of prepaid credit and with no subscription. It needs no batching: the lower price applies to every request, streaming and Claude Code included. On Claude Opus 5.5 it's below even Anthropic's Batch price:
| Model, in / out per 1M | Anthropic list | Anthropic Batch | Zurelay |
|---|---|---|---|
| Claude Opus 5.5 | $4.00 / $20.00 | $2.00 / $10.00 | $1.40 / $7.00 |
| Claude Sonnet 5.5 | $2.00 / $10.00 | $1.00 / $5.00 | $0.79 / $3.95 |
| Claude Haiku 5.5 | $0.10 / $0.50 | $0.05 / $0.25 | $0.04 / $0.20 |
Zurelay doesn't serve Anthropic's Batches or Files APIs, so its price doesn't stack with the Batch discount; for those, or an agreement with Anthropic, go direct. How each route is checked to be the real model is in verification.
Discounts read from Anthropic's pricing docs, plans page and startups program on November 15, 2026.