#
The highest accuracy web search for your AI
Built for frontier teams, trusted by Fortune 500
## A web API purpose-built for AIs
Powering millions of daily requests
### Highest accuracy
Production-ready outputs built on cross-referenced facts, with minimal hallucination.
### Predictable costs
Flex compute budget based on task complexity. Pay per query, not per token.
### Evidence-based outputs
Verifiability and provenance for every atomic output.
### Trusted
SOC-II Type 2 Certified, trusted by leading startups and enterprises.
Highest accuracy at every price point
State of the art across the most challenging benchmarks
COST (CPM)
ACCURACY (%)
CPM: USD per 1000 requests. Cost is shown on a Log scale.
**Dataset**
We evaluated search providers against five open benchmarks covering complementary aspects of agentic search: BrowseComp (hard multi-hop questions that require navigating the live web), Frames (multi-document factoid reasoning), FreshQA (time-sensitive questions where the correct answer depends on recent web information), HLE (Humanity's Last Exam — expert-level academic questions spanning math, science, and humanities), SealQA (ambiguity-robust factoid QA with intentionally misleading snippets), WebWalker (tasks designed around following links across pages to find an answer).
**Evaluation methodology**
Every task is run through a shared deep-research harness: a single GPT-5.4 agent is given two tools (web search and web fetch) with an iterative budget of up to `MAX_TOOL_CALLS=25` tool calls per question. The agent plans sub-queries, fans out searches, fetches specific pages when snippets are insufficient, and returns an answer when it exhausts the number of allowed tool calls or has sufficient information to answer the question. Each answer is then LLM-graded by GPT-5.4. We report accuracy of the final answer.
We measure accuracy and overall cost, which includes LLM token costs and tool call costs.
**Testing dates**
April 19-21, 2026
## Highest accuracy at every price point
State of the art across the most challenging benchmarks
### Humanity's Last Exam
| Series | Model | Cost (CPM) | Accuracy (%) | | -------- | ----------------- | ---------- | ------------ | | Parallel | Parallel Basic | 451 | 58 | | Parallel | Parallel Advanced | 315 | 56 | | Others | Exa | 522 | 57 | | Others | Tavily | 538 | 54 |
CPM: USD per 1000 requests. Cost is shown on a Log scale.
**Dataset**
We evaluated search providers against five open benchmarks covering complementary aspects of agentic search: BrowseComp (hard multi-hop questions that require navigating the live web), Frames (multi-document factoid reasoning), FreshQA (time-sensitive questions where the correct answer depends on recent web information), HLE (Humanity's Last Exam — expert-level academic questions spanning math, science, and humanities), SealQA (ambiguity-robust factoid QA with intentionally misleading snippets), WebWalker (tasks designed around following links across pages to find an answer).
**Evaluation methodology**
Every task is run through a shared deep-research harness: a single GPT-5.4 agent is given two tools (web search and web fetch) with an iterative budget of up to `MAX_TOOL_CALLS=25` tool calls per question. The agent plans sub-queries, fans out searches, fetches specific pages when snippets are insufficient, and returns an answer when it exhausts the number of allowed tool calls or has sufficient information to answer the question. Each answer is then LLM-graded by GPT-5.4. We report accuracy of the final answer.
We measure accuracy and overall cost, which includes LLM token costs and tool call costs.
**Testing dates**
April 19-21, 2026
### FRAMES
| Series | Model | Cost (CPM) | Accuracy (%) | | -------- | ----------------- | ---------- | ------------ | | Parallel | Parallel Advanced | 93 | 87 | | Parallel | Parallel Basic | 165 | 84 | | Others | Exa | 169 | 87 | | Others | Tavily | 189 | 83 |
CPM: USD per 1000 requests. Cost is shown on a Log scale.
**Dataset**
We evaluated search providers against five open benchmarks covering complementary aspects of agentic search: BrowseComp (hard multi-hop questions that require navigating the live web), Frames (multi-document factoid reasoning), FreshQA (time-sensitive questions where the correct answer depends on recent web information), HLE (Humanity's Last Exam — expert-level academic questions spanning math, science, and humanities), SealQA (ambiguity-robust factoid QA with intentionally misleading snippets), WebWalker (tasks designed around following links across pages to find an answer).
**Evaluation methodology**
Every task is run through a shared deep-research harness: a single GPT-5.4 agent is given two tools (web search and web fetch) with an iterative budget of up to `MAX_TOOL_CALLS=25` tool calls per question. The agent plans sub-queries, fans out searches, fetches specific pages when snippets are insufficient, and returns an answer when it exhausts the number of allowed tool calls or has sufficient information to answer the question. Each answer is then LLM-graded by GPT-5.4. We report accuracy of the final answer.
We measure accuracy and overall cost, which includes LLM token costs and tool call costs.
**Testing dates**
April 19-21, 2026
### WebWalker
| Series | Model | Cost (CPM) | Accuracy (%) | | -------- | ----------------- | ---------- | ------------ | | Others | Exa | 210 | 74 | | Others | Tavily | 202 | 71 | | Parallel | Parallel Advanced | 101 | 73 | | Parallel | Parallel Basic | 155 | 71 |
CPM: USD per 1000 requests. Cost is shown on a Log scale.
**Dataset**
We evaluated search providers against five open benchmarks covering complementary aspects of agentic search: BrowseComp (hard multi-hop questions that require navigating the live web), Frames (multi-document factoid reasoning), FreshQA (time-sensitive questions where the correct answer depends on recent web information), HLE (Humanity's Last Exam — expert-level academic questions spanning math, science, and humanities), SealQA (ambiguity-robust factoid QA with intentionally misleading snippets), WebWalker (tasks designed around following links across pages to find an answer).
**Evaluation methodology**
Every task is run through a shared deep-research harness: a single GPT-5.4 agent is given two tools (web search and web fetch) with an iterative budget of up to `MAX_TOOL_CALLS=25` tool calls per question. The agent plans sub-queries, fans out searches, fetches specific pages when snippets are insufficient, and returns an answer when it exhausts the number of allowed tool calls or has sufficient information to answer the question. Each answer is then LLM-graded by GPT-5.4. We report accuracy of the final answer.
We measure accuracy and overall cost, which includes LLM token costs and tool call costs.
**Testing dates**
April 19-21, 2026
### FreshQA
| Series | Model | Cost (CPM) | Accuracy (%) | | -------- | ----------------- | ---------- | ------------ | | Parallel | Parallel Advanced | 49 | 79 | | Parallel | Parallel Basic | 90 | 77 | | Others | Exa | 84 | 78 | | Others | Tavily | 89 | 78 |
CPM: USD per 1000 requests. Cost is shown on a Log scale.
**Dataset**
We evaluated search providers against five open benchmarks covering complementary aspects of agentic search: BrowseComp (hard multi-hop questions that require navigating the live web), Frames (multi-document factoid reasoning), FreshQA (time-sensitive questions where the correct answer depends on recent web information), HLE (Humanity's Last Exam — expert-level academic questions spanning math, science, and humanities), SealQA (ambiguity-robust factoid QA with intentionally misleading snippets), WebWalker (tasks designed around following links across pages to find an answer).
**Evaluation methodology**
Every task is run through a shared deep-research harness: a single GPT-5.4 agent is given two tools (web search and web fetch) with an iterative budget of up to `MAX_TOOL_CALLS=25` tool calls per question. The agent plans sub-queries, fans out searches, fetches specific pages when snippets are insufficient, and returns an answer when it exhausts the number of allowed tool calls or has sufficient information to answer the question. Each answer is then LLM-graded by GPT-5.4. We report accuracy of the final answer.
We measure accuracy and overall cost, which includes LLM token costs and tool call costs.
**Testing dates**
April 19-21, 2026
### Coding
| Series | Model | Cost (CPM) | Accuracy (%) | | -------- | ----------------- | ---------- | ------------ | | Parallel | Parallel Advanced | 154 | 82 | | Parallel | Parallel Basic | 269 | 81 | | Others | Exa | 331 | 80 | | Others | Tavily | 352 | 75 |
CPM: USD per 1000 requests. Cost is shown on a Log scale.
**Dataset**
A proprietary coding dataset derived from production queries to Parallel’s search API.
**Evaluation methodology**
Every task is run through a shared deep-research harness: a single GPT-5.4 agent is given two tools (web search and web fetch) with an iterative budget of up to `MAX_TOOL_CALLS=25` tool calls per question. The agent plans sub-queries, fans out searches, fetches specific pages when snippets are insufficient, and returns an answer when it exhausts the number of allowed tool calls or has sufficient information to answer the question. Each answer is then LLM-graded by GPT-5.4. We report the accuracy of the final answer.
We measure accuracy and overall cost, which includes LLM token costs and tool call costs.
**Testing dates**
April 19-21, 2026
Search, built for AIs
The most accurate search tool, to bring web context to your AI agents
The most accurate deep and wide research
Run deeper and more accurate research at scale, for the same compute budget
Build a dataset from the web
Define your search criteria in natural language, and get back a structured table of matches
Custom web enrichment
Bring existing data, define output columns to research, and get fresh web enrichments back
Monitor any event on the web
Continuously monitor for any changes on the web
Start building
## Towards a programmatic web for AIs
Parallel is building new interfaces, infrastructure, and business models for AIs to work with the web
## Agent onboarding prompt
Use curl to read [parallel.ai/agents.md](https://parallel.ai/agents.md) and perform the setup to install Parallel
## Agent onboarding prompt
Use curl to read parallel.ai/agents.md and perform the setup to install Parallel
Latest updates