Anthropic / Claude API Pricing — KeyoAPI vs Typical List Rates

Looking up Anthropic API pricing or Claude API pricing usually means you already want Anthropic-class chat or agents, but need clearer unit economics — or one prepaid balance that also covers GPT-class, DeepSeek, Whisper, and vision. This page compares indicative Keyo sell rates with typical public Anthropic list figures and shows how to call Claude-class IDs on an OpenAI-compatible gateway.

Anthropic / Claude API Pricing is a model-family pricing lane on KeyoAPI — a prepaid AI API gateway where one Bearer key and one wallet ($1 = 500,000 quota units) cover chat, speech, vision, OCR and video models.

Price comparison table

Indicative figures for planning. Confirm live Keyo sell rates on interactive Model Square or the static pricing list.

Prototype on gift-credit models, then promote to Claude

Keyo does not ship Anthropic-branded free Claude IDs. To prototype on the same key, use gift-credit catalog IDs on the gift-credit models page — gift-credit IDs that bill at dedicated gift-credit rates from your one-time $10 new-user credit (fair-use rate limits). When quality gates pass, switch model= to claude-sonnet-5 or another Claude ID without changing base URL or auth.

Cheapest Claude API rate on Keyo

The cheapest Claude-class lane we list is claude-haiku-5-5 at ~$0.07 / $0.37 per 1M in/out — the fastest, cheapest Claude 5.5 lane — and it is the row to start on if you searched for a cheap Claude API. For balanced coding work, claude-sonnet-5 (~$0.44 / $2.21) is the everyday lane.

Above it, by task rather than by tier: claude-opus-5 ~$1.10 / $5.52 for hard agent and coding escalations — the current Opus generation, claude-opus-5-5, sits at ~$1.76 / $8.82 for the hardest agent loops; claude-fable-5 ~$4.41 / $22.06 and claude-fable-5-1 ~$7.35 / $36.77 for flagship chat. Cheaper than all of them and still on the same key: gpt-6-luna (~$0.03 / $0.15) for high-volume drafting, or deepseek-v4.1-flash-1017 (~$0.02 / $0.06) if the task does not actually need Claude reasoning.

Cost math: 10M input + 2M output tokens/month on claude-sonnet-5 ≈ $4.40 + $4.42 ≈ $8.82. The same mix on claude-fable-5 ≈ $44.12 + $44.12 ≈ $88.24. Confirm live rates on the claude-sonnet-5 model page before you commit volume.

Anthropic API pricing at a glance

Anthropic API pricing at a glance: the table above lists indicative Keyo sell rates for Claude-class IDs next to typical Anthropic public list bands. Teams comparing Anthropic API pricing (and Claude API pricing) usually already want Anthropic-class chat or agents with clearer unit economics — or one prepaid balance across Claude-class IDs plus GPT-class, DeepSeek, Whisper, OCR, and TTS. Official Anthropic surfaces win on brand trust; KeyoAPI wins when you want that stack without five invoices.

Keyo rates above are indicative sell prices from the live Keyo catalog. Always confirm interactive pages under the claude-sonnet-5 model page, the claude-opus-5 model page, the claude-opus-5-5 model page, the claude-fable-5 model page, and the claude-fable-5-1 model page before contracting volume. Anthropic list bands and context tiers change — treat official figures as planning context. For the per-token math behind these rows — what one request actually costs and how to sanity-check it against real traffic — see the guide: our blog guide.

How to call: set OPENAI_BASE_URL=https://www.keyoapi.xyz/v1, use your Keyo API key, and pass a Claude model string (for example claude-sonnet-5) to POST /v1/chat/completions. Longer guides live at the claude-sonnet-5 model page, the claude-opus-5 model page, the claude-opus-5-5 model page, and the Fable pages.

When Claude returns HTTP 529 overloaded (or you are planning retries around Claude API overloaded errors), see the developer playbook: our blog guide — retry/backoff and fallback patterns on the same Keyo key.

When Anthropic / Claude list prices spike your margin, keep Claude for hard reasoning and shift bulk classification or drafts to gpt-6-luna or deepseek-v4.1-flash-1017 on the same key. For OpenAI / GPT dollars see the OpenAI API pricing page; for Gemini dollars see the Gemini API pricing page; for DeepSeek see the DeepSeek API pricing page; for Grok see the Grok API pricing page; for free prototyping see the gift-credit models page.

Opus 5.5 vs Fable 5.1

At the default-group rates checked on 2026-10-13, claude-opus-5-5 lists $1.7647 / $8.8235 per million input/output tokens and claude-fable-5-1 lists $7.3530 / $36.7650. Fable tokens cost about 4.17 times as much. See Fable 5.1 vs Opus 5.5 for the request-cost calculation and coding or agent evaluation checklist. The Opus model guide is the claude-opus-5-5 model page.

Claude Code API pricing: budget the full task

See Claude Code API Pricing: Budget the Whole Task for exact default-group rates and a worked 100K input / 10K output task budget. The token mix is an explicit assumption, not an average Claude Code session. Compare current subscription terms separately, and verify the supported gateway format before connecting a client.

A separate decision model for routing

For bounded routing, the cloudflare-clef model page is a Cloudflare decision model with POST /v1/decisions. It returns typed probabilities and is not part of the Claude chat price table. An application can use it to classify a request before sending selected work to a chat model; evaluate that routing policy on your own tasks.

curl https://www.keyoapi.xyz/v1/chat/completions \
  -H "Authorization: Bearer $KEYO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"claude-haiku-5-5","messages":[{"role":"user","content":"Hello"}]}'

Get started in five steps

1. Create an account — new users receive $10 of gift credit for prototyping.
2. Create an API key in the console.
3. Point your OpenAI SDK at https://www.keyoapi.xyz/v1 with the model ID from this page, or call the dedicated REST path from the API handbook.
4. Track spend in the console Wallet — $1 equals 500,000 quota units, and gift credit burns before recharged balance.
5. Set a per-key spending cap before wiring production traffic.

A verified first call is the acceptance test for the setup: the response should carry the payload shape documented in the handbook, the Wallet ledger should show a deduction in quota units (500,000 units = $1), and repeated calls should deduct linearly at the listed rate. Any mismatch usually means the model ID or endpoint is wrong — check both against the pricing list before debugging client code.

FAQ

Also see the model comparison page, the gift-credit models page, the full pricing list, and claude-sonnet-5 · claude-opus-5.