Every Claude model — specs, pricing, context window, capabilities, and best use cases.
Anthropic was founded in 2021 by Dario Amodei, Daniela Amodei, and other former OpenAI researchers. The company focuses on AI safety research and deploys AI through its Claude model family. Anthropic is known in the market for models with strong writing quality, nuanced instruction-following, large context windows, and a safety-focused training methodology (Constitutional AI). All Claude models support vision (image input), tool use / function calling, and are accessible via API and the Claude.ai interface.
---
| Property | Value | |---|---| | API Identifier | claude-3-haiku-20240307 | | Released | March 2024 | | Context Window | 200,000 tokens (~150,000 words) | | Input Pricing | $0.25 per million tokens | | Output Pricing | $1.25 per million tokens | | Vision | Yes | | Tool Use | Yes |
Best for: High-volume, latency-sensitive workloads. Customer support triage, content classification, summarization pipelines, lightweight chat. Anywhere cost and speed matter more than peak quality.
Strengths: Fastest model in the Claude 3 family, lowest cost, reliable for well-defined tasks.
Weaknesses: Noticeably weaker on complex reasoning, nuanced writing, and multi-step tasks compared to larger Claude models.
---
| Property | Value | |---|---| | API Identifier | claude-3-sonnet-20240229 | | Released | March 2024 | | Context Window | 200,000 tokens | | Input Pricing | $3.00 per million tokens | | Output Pricing | $15.00 per million tokens |
Status: Superseded by Claude 3.5 Sonnet. Still available via API but no longer recommended for new projects. At the same price, Claude 3.5 Sonnet performs substantially better.
---
| Property | Value | |---|---| | API Identifier | claude-3-opus-20240229 | | Released | March 2024 | | Context Window | 200,000 tokens | | Input Pricing | $15.00 per million tokens | | Output Pricing | $75.00 per million tokens |
Status: Was Anthropic's flagship model at launch, positioned as the most capable Claude at the time. Superseded by Claude 3.5 Sonnet, which performs comparably or better at one-fifth the price. Still available; may be preferred by users who built workflows around it.
Strengths: Strong overall performance across tasks, nuanced instruction-following.
Weaknesses: Very expensive relative to performance; newer models have surpassed it.
---
| Property | Value | |---|---| | API Identifier | claude-3-5-sonnet-20241022 | | Released | October 2024 (updated from June 2024 original) | | Context Window | 200,000 tokens | | Input Pricing | $3.00 per million tokens | | Output Pricing | $15.00 per million tokens | | Vision | Yes | | Tool Use | Yes | | Computer Use | Yes (beta) |
Best for: General-purpose flagship use. Complex reasoning, coding, writing, analysis, agentic tasks. The recommended default for most developers who need high quality.
Strengths: Anthropic's best overall model as of late 2024 on most standard benchmarks. Excellent at coding (strong HumanEval performance), nuanced writing, and long-document analysis. Computer Use capability allows the model to interact with GUIs (in beta). Strong balance of cost vs. capability.
Weaknesses: Does not perform extended reasoning (chain-of-thought with longer computation budget) — that is Claude 3.7 Sonnet's domain. At $3/$15 per million tokens it is not cheap for very high-volume use.
---
| Property | Value | |---|---| | API Identifier | claude-3-5-haiku-20241105 | | Released | November 2024 | | Context Window | 200,000 tokens | | Input Pricing | $0.80 per million tokens | | Output Pricing | $4.00 per million tokens | | Vision | Yes | | Tool Use | Yes |
Best for: Applications that need more capability than Claude 3 Haiku but better economics than Claude 3.5 Sonnet. Customer-facing chat, content processing, internal tools.
Strengths: Matches or exceeds Claude 3 Opus on several benchmarks at a fraction of the cost. Good balance of speed, capability, and price. Significantly smarter than Claude 3 Haiku.
Weaknesses: Not as capable as Claude 3.5 Sonnet or 3.7 Sonnet for the most demanding tasks.
---
| Property | Value | |---|---| | API Identifier | claude-3-7-sonnet-20250219 | | Released | February 2025 | | Context Window | 200,000 tokens | | Input Pricing | $3.00 per million tokens | | Output Pricing | $15.00 per million tokens | | Vision | Yes | | Tool Use | Yes | | Extended Thinking | Yes (optional, additional token cost) |
Best for: Tasks requiring deep reasoning, complex mathematics, hard coding problems, and multi-step analysis where higher accuracy justifies longer processing time.
Strengths: Anthropic's most capable model as of early 2025. Extended thinking mode allocates additional computation budget to reasoning before producing a final answer — visible as thinking tokens in the API response. Particularly strong on complex coding and mathematical reasoning benchmarks. Can be used in standard mode (no extended thinking) for general tasks.
Weaknesses: Extended thinking mode increases latency and cost (thinking tokens are also billed). Not necessary for most everyday tasks. Standard mode performance is comparable to 3.5 Sonnet.
---
| Plan | Price | Key Features | |---|---|---| | Free | $0/month | Access to Claude, limited usage, Claude 3.5 Haiku primarily | | Pro | $20/month | Higher usage limits, access to all models including Claude 3.7 Sonnet, Projects feature, priority access | | Team | $25/user/month (min 5 users) | All Pro features, shared Projects, admin console, higher rate limits | | Enterprise | Custom pricing | SSO, expanded context, advanced admin, security and compliance features, dedicated support |
---
Known for: Writing quality (many users rate Claude as the best model for nuanced long-form writing), strong instruction-following, large context windows (200K across all models), refusal calibration (considered by many to be better-calibrated than competitors — less prone to refusing legitimate requests while still declining harmful ones), and safety research transparency.
API ecosystem: Anthropic's API is widely supported through major cloud providers including AWS (Amazon Bedrock) and Google Cloud (Vertex AI).
Pricing page: anthropic.com/pricing
Have a follow-up question about this topic?
Ask AI