tavily-web
Verifiedby sickn33 · Updated Apr 16, 2026
Web search, content extraction, crawling, and research capabilities using Tavily API. Use when you need to search the web for current information, extracting content from URLs, or crawling websites.
See It In Action
Let me pull the latest pricing from all three providers' official docs simultaneously.
Got Gemini pricing. Let me fetch the other two from alternative URLs.
OpenAI's pricing page is blocking direct fetch. Let me search for it.
Let me get more complete OpenAI pricing data.
Got everything. Here's the comparison:
API Pricing & Limits Comparison (April 2026)
All prices per 1M tokens in USD.
Flagship / High-Intelligence Models
| Model | Input | Output | Cached Input | Context | Max Output |
|---|---|---|---|---|---|
| OpenAI GPT-5.4 | $2.50 | $15.00 | $0.25 | — | — |
| OpenAI GPT-5.4-pro | $30.00 | $180.00 | — | — | — |
| Anthropic Opus 4.6 | $5.00 | $25.00 | $0.50 | 1M | 128k |
| Google Gemini 3.1 Pro (preview) | $2–4 | $12–18 | — | — | — |
| Google Gemini 2.5 Pro | $1.25 | $10.00 | $0.125 | 1M+ | — |
Mid-Tier / Balanced Models
| Model | Input | Output | Cached Input | Context | Max Output |
|---|---|---|---|---|---|
| OpenAI GPT-5.4-mini | $0.75 | $4.50 | $0.075 | — | — |
| Anthropic Sonnet 4.6 | $3.00 | $15.00 | $0.30 | 1M | 64k |
| Google Gemini 3 Flash (preview) | $0.50 | $3.00 | — | — | — |
| Google Gemini 2.5 Flash | $0.30 | $2.50 | $0.03 | 1M | — |
Budget / Fast Models
| Model | Input | Output | Cached Input | Context | Max Output |
|---|---|---|---|---|---|
| OpenAI GPT-5.4-nano | $0.20 | $1.25 | $0.02 | — | — |
| Anthropic Haiku 4.5 | $1.00 | $5.00 | $0.10 | 200k | 64k |
| Google Gemini 2.5 Flash-Lite | $0.10 | $0.40 | — | — | — |
Legacy OpenAI (still available, off main pricing page)
| Model | Input | Output |
|---|---|---|
| GPT-4.1 | $2.00 | $8.00 |
| GPT-4o | $2.50 | $10.00 |
| o3 | $2.00 | $8.00 |
| o4-mini | $1.10 | $4.40 |
Batch API Discounts
| Provider | Discount |
|---|---|
| OpenAI | 50% off (async, 24h turnaround) |
| Anthropic | 50% off (async) |
| 50% off |
Prompt Caching
| Provider | Savings on cache hits |
|---|---|
| OpenAI | 90% off (GPT-5.x), 75% off (GPT-4.1), 50% off (GPT-4o/o-series) |
| Anthropic | 90% off base input (all models); write costs 1.25x–2x depending on TTL |
| Varies by model; $0.03–$0.25/MTok read + storage fees |
Key Takeaways
- Cheapest capable model: Gemini 2.5 Flash-Lite ($0.10/$0.40) or GPT-5.4-nano ($0.20/$1.25)
- Best value at the top: Gemini 2.5 Pro is very competitive at $1.25/$10 with 1M context
- Anthropic premium: Opus 4.6 is the priciest flagship ($5/$25) but has confirmed 1M context + 128k output
- Google free tier: Gemini Flash models have a free tier with rate limits — useful for prototyping
- OpenAI context: GPT-5.4 series context windows weren't listed on the pricing page; GPT-4.1 had 1M
Note: OpenAI's pricing page blocked direct scraping and some GPT-5.4 details (context window, max output, rate limits) weren't fully exposed. I'd recommend double-checking those on their docs directly.
Sources:
Permissions
| Scope | Description |
|---|---|
| network:outbound | |
| env:read |
SKILL.md
tavily-web
Overview
Web search, content extraction, crawling, and research capabilities using Tavily API
When to Use
- When you need to search the web for current information
- When extracting content from URLs
- When crawling websites
Installation
npx skills add -g BenedictKing/tavily-web
Step-by-Step Guide
- Install the skill using the command above
- Configure Tavily API key
- Use naturally in Claude Code conversations
Examples
See GitHub Repository for examples.
Best Practices
- Configure API keys via environment variables
Troubleshooting
See the GitHub repository for troubleshooting guides.
Related Skills
- context7-auto-research, exa-search, firecrawl-scraper, codex-review
FAQ
What does tavily-web do?
Web search, content extraction, crawling, and research capabilities using Tavily API. Use when you need to search the web for current information, extracting content from URLs, or crawling websites.
When should I use tavily-web?
Use it when you need a repeatable workflow that produces text response.
What does tavily-web output?
In the evaluated run it produced text response.
How do I install or invoke tavily-web?
Ask the agent to use this skill when the task matches its documented workflow.
Which agents does tavily-web support?
Agent support is inferred from the source, but not explicitly declared.
What tools, channels, or permissions does tavily-web need?
It uses no extra tools; channels commonly include text; permissions include network:outbound, env:read.
Is tavily-web safe to install?
Static analysis marked this skill as medium risk; review side effects and permissions before enabling it.
How is tavily-web different from an MCP or plugin?
A skill packages instructions and workflow conventions; tools, MCP servers, and plugins are dependencies the skill may call during execution.
Does tavily-web outperform not using a skill?
About tavily-web
When to use tavily-web
You need up-to-date information from the public web. You want to extract readable content from one or more URLs. You need to crawl a website as part of a research task.
When tavily-web is not the right choice
You only need local file analysis with no web access. You do not have a Tavily API key configured.
What it produces
Produces text response.