Anthropic sells Claude through a pay-as-you-go API with four model lines: Fable at the top, then Opus, Sonnet and Haiku. The rates below come from our dataset, which tracks Anthropic's pricing page. The Claude.ai Free, Pro and Max plans are a separate product.
Claude Haiku 5.5 is the cheapest Claude model at $0.10 input / $0.50 output per 1M tokens for prompts up to 100K tokens. Sonnet 5.5 costs $2.00 / $10.00, Opus 5.5 $4.00 / $20.00 and Fable 5.1 $10.00 / $50.00. The Batch API halves both input and output prices.
| Model | Tier | Input / 1M | Output / 1M | Cache hit / 1M | Context | Notes |
|---|---|---|---|---|---|---|
| Claude Fable 5.1 | Flagship | $10.00 | $50.00 | $0.25 | 1M | Most powerful; released Sep 1, 2026, replacing Fable 5 at same per-token price (cache reads cut 75% to $0.25) |
| Claude Opus 5 | Flagship | $5.00 | $25.00 | $0.50 | 1M | New Opus flagship (Jul 24); near-Fable 5 at half the price |
| Claude Opus 5.5 | Flagship | $4.00 | $20.00 | $0.20 | 1M | New leading Opus model (Sep 22, 2026); Fable 5.1-level performance at 20% less than Opus 5; included on Pro/Max/Team at no extra cost |
| Claude Sonnet 5.5 | Mid-tier | $2.00 | $10.00 | $0.10 | 1M | Released Sep 28, 2026, replacing Sonnet 5 at the identical $2/$10 per 1M price; Sonnet 5 is now a legacy model |
| Claude Haiku 5.5 | Budget | $0.10 | $0.50 | $0.01 | 1M | Released Oct 7, 2026; 90% cheaper than Haiku 4.5 with a 1M context window (for prompts ≤100K tokens); >100K tokens: $0.50/$2.50, cache read $0.05, cache write $0.625 |
| Claude Haiku 4.5 | Budget | $1.00 | $5.00 | $0.10 | 200K | Fast, near Sonnet 4 quality; now a legacy model, superseded by Haiku 5.5 (Oct 7, 2026) but still available |
| Model | Support chatbot 10,000 conversations × (2,000 input + 400 output tokens) | Document Q&A 1,000 questions × (30,000 input + 800 output tokens) | Coding agent session 4M input tokens (90% cache hits) + 150,000 output tokens |
|---|---|---|---|
| Claude Haiku 5.5 | $4.00 | $3.40 | $0.15 |
| Claude Haiku 4.5 | $40.00 | $34.00 | $1.51 |
| Claude Sonnet 5.5 | $80.00 | $68.00 | $2.66 |
| Claude Opus 5.5 | $160 | $136 | $5.32 |
| Claude Opus 5 | $200 | $170 | $7.55 |
| Claude Fable 5.1 | $400 | $340 | $12.40 |
Standard rates per 1M tokens from data on this site. The coding-agent row assumes 90% of input is read from the prompt cache (full input price where a model has no cache rate). Cache-write fees, batch discounts and long-context surcharges listed in the notes are not included. Updated November 2, 2026.
| Model | Provider | Tier | Input / 1M | Output / 1M | Context |
|---|---|---|---|---|---|
| Claude Opus 5.5 | Anthropic | Flagship | $4.00 | $20.00 | 1M |
| Grok 4.7 | SpaceXAI | Flagship | $2.00 | $6.00 | 500K |
| Qwen3.8-Max | Alibaba | Flagship | $2.00 | $6.00 | 1M |
| Gemini 4 Argon | Flagship | $2.00 | $10.00 | 1M | |
| Kimi K3.1 Pro | Moonshot AI | Flagship | $2.00 | $10.00 | 1M |
| Claude Sonnet 5.5 | Anthropic | Mid-tier | $2.00 | $10.00 | 1M |
| Mistral Large 3 | Mistral | Mid-tier | $0.50 | $1.50 | 256K |
| DeepSeek V4-Pro | DeepSeek | Mid-tier | $0.66 | $1.98 | 1M |
| Mistral Large 4 | Mistral | Mid-tier | $0.70 | $2.10 | 1M |
| Grok 4.3 | SpaceXAI | Mid-tier | $1.25 | $2.50 | 1M |
| Claude Haiku 5.5 | Anthropic | Budget | $0.10 | $0.50 | 1M |
| Ministral 3 14B | Mistral | Budget | $0.20 | $0.20 | 256K |
| Gemini 2.5 Flash-Lite | Budget | $0.10 | $0.40 | 1M | |
| GPT-6 Luna | OpenAI | Budget | $0.10 | $0.50 | 1M |
| Mistral Small 4 | Mistral | Budget | $0.15 | $0.60 | 256K |
Anthropic's cheapest model per tier (bold) against the four cheapest same-tier models from other providers, by output price.
Claude Opus 5.5 costs $4.00 / $20.00 against Opus 5's $5.00 / $25.00, and Anthropic bills its cache hits at 0.05× the input price instead of the usual 0.1×. Sonnet 5.5 gets the same 0.05× rate, and Fable 5.1 0.025×. Unless you have pinned prompts tuned for Opus 5, switching is a straight saving. The same goes for the budget tier: Haiku 5.5 costs a tenth of legacy Haiku 4.5 ($1.00 / $5.00) and has a 1M window against 200K.
A 5-minute cache write costs 1.25× the base input price and a 1-hour write 2×. On Sonnet 5.5 that means $2.50 to write a prefix and $0.10 to read it back, against $2.00 uncached. A 5-minute write pays for itself after one cache read, and a 1-hour write after two. Cache reads also don't count toward your input-tokens-per-minute rate limit on current models. A well-cached agent therefore gets more throughput as well as a smaller bill. Put the system prompt, tool definitions and large documents first, and the changing user turn last.
Opus 5.5, Sonnet 5.5 and Fable 5.1 bill their full 1M-token window at the standard per-token rate: a 900K-token request pays the same rate as a 9K-token one. Haiku 5.5 is the exception. A request whose prompt goes over 100,000 tokens (cache reads and writes included) pays 5× the base rates for that whole request. Even then Haiku 5.5 stays well below Sonnet 5.5, but if your typical document is 150K tokens, compare the two before assuming the small model is a bargain.
Three things move a Claude bill away from the list price. First, Claude 4.7 and later models use a newer tokenizer that Anthropic says produces about 30% more tokens for the same text than earlier Claude models, so compare costs per task, not just per token. Second, requesting US-only inference (inference_geo: "us") adds a 1.1× multiplier to every token category. Third, fast mode on Opus 5.5 doubles its input and output rates and can't be combined with the Batch API. Server tools such as web search are also charged per use on top of tokens.
Not an ongoing one. The Claude API is pay-as-you-go; Anthropic's pricing page says only that new users receive a small amount of free credits to test the API, and new organizations may start in an Evaluation tier with lower rate limits. Claude Max, Ultra and Team subscribers can also claim monthly API credits by linking a Claude Console organization (Free, Pro and Enterprise plans aren't eligible). If you need a lasting free allowance to prototype, Google's Gemini API has a free tier on most Flash models.
Claude Haiku 5.5 costs $0.10 per 1M input tokens and $0.50 per 1M output tokens, with cache hits at $0.01, for prompts up to 100K tokens. Longer prompts pay 5× those rates. The legacy Claude Haiku 4.5 is still available at $1.00 / $5.00. Batch requests cost half on both.
Tier for tier, the list prices are close: Haiku 5.5 vs GPT-6 Luna ($0.10 / $0.50), Sonnet 5.5 vs GPT-6.1 Sol ($2.00 / $10.00). At the top, Opus 5.5 ($20.00 output) undercuts GPT-6 Astra ($50.00). Claude is cheaper on prompts beyond 272K tokens, where OpenAI raises its rates and Sonnet and Opus don't. Below that, compare cost per task rather than per token: Anthropic says its current tokenizer produces about 30% more tokens for the same text than its earlier one, and OpenAI's tokenizer counts differently again.
Claude Pro does not. Claude Max ($100–200/mo), Ultra and Team plans include a monthly API credit: after 7 days on the plan, you claim it in claude.ai billing settings by linking a Claude Console organization. Unused credit expires at the end of each billing cycle, it can't be spent on Claude Code, and it doesn't change your usage limits in the Claude apps.
For most workloads, test Opus 5.5 first. Fable 5.1 costs $10.00 / $50.00, while Anthropic positions Opus 5.5 at Fable 5.1-level performance for $4.00 / $20.00. Fable 5.1's cache hits are the cheapest relative to input ($0.25), which narrows the gap only on heavily cached agent loops.
How we choose: prices come from each vendor's official pricing page and are re-checked every week. We take no affiliate fees, so rankings aren't for sale.