Best search accuracy
Exa leads
- Exa fast
- 99.3%
- Tavily advanced
- 93.0%
Correct answers on the same search questions.
independent benchmark · same questions · official pricing
Exa vs Tavily: Web Search API Comparison, Pricing & Benchmark (2026)
Exa vs Tavily for AI apps, agents and RAG: compare search accuracy, speed and cost per 1,000 queries. Best accuracy: Exa fast 99.3% vs Tavily advanced 93.0%. Fastest modes: Exa instant 398ms vs Tavily basic 1.88s. Lowest tested pay-as-you-go price: Exa $7.00 vs Tavily $8.00 per 1,000 queries. Independent benchmark on 300 questions, with official docs and pricing. Evaluate Exa as a Tavily alternative, with context on Nebius’s Tavily acquisition.
Compared on the same 300 search questions. Each metric below names the mode it measures.
Fastest search
Exa leads
- Exa instant
- 398ms
- Tavily basic
- 1.88s
Mean time for a search request.
Lowest price / 1,000 queries
Exa leads
- Exa fast
- $7.00
- Tavily basic
- $8.00
Tested modes, pay-as-you-go list prices.
Pricing depends on the plan: Tavily’s prepaid rates can be lower than its pay-as-you-go rate. See the price comparison and volume example →
Tavily is now part of Nebius. Nebius acquired Tavily on 19 February 2026. Comparing Tavily alternatives? Start with Exa’s measured search results above, then review the acquisition and API continuity →
Exa vs Tavily: Web Search API Comparison Benchmarked
All tested search modes, ranked by accuracy. Speed is mean search latency; cost is the API list price per 1,000 queries.
| Search mode | Accuracy | Search time | Cost / 1,000 queries | Official docs & pricing |
|---|---|---|---|---|
| Exa fasttype=fast | 99.3% | 652ms | $7.00pay as you go | Exa official docs ↗ Official pricing ↗ |
| Exa instanttype=instant | 97.7% | 398ms | $7.00pay as you go | Exa official docs ↗ Official pricing ↗ |
| Tavily advancedsearch_depth=advanced | 93.0% | 4.29s | $16.00pay as you go | Tavily official docs ↗ Official pricing ↗ |
| Tavily basicsearch_depth=basic | 87.7% | 1.88s | $8.00pay as you go | Tavily official docs ↗ Official pricing ↗ |
Full search metrics: evidence recall, tokens and cost per correct answer
| Rank | Vendor | Endpoint & configuration | Official docs | Accuracy | AR@1 | AR@5 | $ / 1k correct | Latency | Snippet tokens | Total $ | Cost per 1,000 queries |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Exaweb search | POST /searchtype=fast | Exa official docs ↗ | 99.3% | 95.0% | 99.3% | $7.05 | 652ms | 1,987 | $2.10 | $7 / 1ktype=fast · up to 10 results |
| 2 | Exaweb search | POST /searchtype=instant | Exa official docs ↗ | 97.7% | 80.0% | 97.3% | $7.17 | 398ms | 2,128 | $2.10 | $7 / 1ktype=instant · up to 10 results |
| 3 | Tavilyweb search | POST /searchsearch_depth=advanced | Tavily official docs ↗ | 93.0% | 65.0% | 92.7% | $17.20 | 4.29s | 2,210 | $4.80 | $16 / 1k2 credits · $0.008 PAYG |
| 4 | Tavilyweb search | POST /searchsearch_depth=basic | Tavily official docs ↗ | 87.7% | 82.0% | 87.7% | $9.13 | 1.88s | 1,639 | $2.40 | $8 / 1k1 credit · $0.008 PAYG |
$ / 1k correct is $ / 1k queries divided by extracted-answer accuracy. Ranked by accuracy, highest first. Cost per 1,000 queries is the published PAYG list price, linked to the vendor pricing page. Not promotional packs or volume discounts. TinyFish Search is free with a 30 requests/min cap — $0 is not unlimited throughput.
Exa vs Tavily Pricing per 1,000 Searches
For the search modes above: Exa instant / fast $7.00, Tavily basic $8.00, Tavily advanced $16.00 per 1,000 queries on pay-as-you-go rates. Prices checked 19 September 2026.
Exa: request-based pricing
The listed search price includes up to 10 results. Additional results above 10 add $1 per 1,000 requests per extra result; AI summaries have separate charges. Deep search has its own rate.
Free credits: $20 at signup, then $10 monthly.
Exa official pricing ↗Tavily: credit-based pricing
Basic uses one credit per request; advanced uses two. Pay-as-you-go credits cost $0.008. Prepaid monthly plans offer lower unit rates when used fully.
Free credits: 1,000 per month.
Tavily official pricing ↗What would 100,000 searches cost?
| Mode / billing | Search API cost | Basis |
|---|---|---|
| Exa instant / fast, pay as you go | $700.00 | 100 × $7.00 per 1,000 |
| Tavily basic, pay as you go | $800.00 | 100 × $8.00 per 1,000 |
| Tavily advanced, pay as you go | $1600.00 | 100 × $16.00 per 1,000 |
| Tavily basic, Growth monthly plan | $500.00 | 100,000 included credits, fully used |
The Growth example makes Tavily basic $5 per 1,000 searches at full use. Compare your monthly volume and plan commitment. LLM generation and additional scrape calls are separate application costs.
Nebius Acquired Tavily: What It Means for Search API Users
Nebius owns Tavily. The agreement was announced on 10 February 2026; Nebius’s financial filing records the completed acquisition on 19 February 2026. Its 20 May update confirms that Tavily continues serving developers under the Tavily brand.
Existing Tavily integrations
Tavily said its API and customer data policies would remain unchanged when it joined Nebius, and reaffirmed its zero-data-retention commitment. Its search API remains available through the official Tavily docs.
Tavily’s customer announcement ↗Choosing your search API now
Use current search accuracy, latency, returned content and your expected bill to choose between Tavily and Exa. The prices above were checked on 19 September 2026, after the acquisition. The benchmark compares API performance; ownership is separate buying context.
Compare current pricing per 1,000 searches →Tavily Alternatives: Exa for AI Search, Agents and RAG
Exa is a Tavily alternative for web search. In the primary benchmark, Exa fast scores 99.3% accuracy versus 93.0% for Tavily advanced. The fastest tested Exa mode is Exa instant at 398ms; Tavily’s is Tavily basic at 1.88s. Exa instant and fast cost $7.00 per 1,000 queries; Tavily basic costs $8.00 and advanced $16.00 on pay-as-you-go rates. Tavily’s Growth plan can lower basic search to $5 per 1,000 at full use.
Test Exa as your Tavily alternative
Run the same queries through both APIs. Compare useful evidence, mean search time and cost per 1,000 requests. Match Tavily’s search_depth to an Exa type, then map the returned URLs and content fields to your application. Verify any scrape or crawl workflow separately using each provider’s docs.
Compare API settings and official docs →Compare more Tavily alternatives
Broaden the shortlist to Brave, Parallel, Firecrawl and the other providers in the full benchmark. Compare each tested mode on the same accuracy, speed and cost basis.
All web search API alternatives →Search Features, Returned Content and Official Docs
Choose the API settings that match your AI app, agent or RAG pipeline. Capability descriptions below come from official docs; accuracy and latency above come from our benchmark.
| What to compare | Exa | Tavily |
|---|---|---|
| Search endpoint | POST /search | POST /search |
| Modes in the primary comparison | instant, fast | basic, advanced |
| Choose a mode | type selects speed and search depth. | search_depth controls the retrieval mode and credit cost. |
| Content for the LLM | Source URLs, text and highlights. | Source URLs and result content; optional cleaned raw content. |
| Control response size | Set numResults and content options. | Set max_results and include_raw_content. |
| Docs | Exa official search docs ↗ | Tavily official search docs ↗ |
Choosing for accuracy and speed
Start with Exa fast and Tavily advanced for each vendor’s highest measured accuracy. For a faster user-facing response, compare Exa instant with Tavily basic. Check the score and latency together for the exact mode you plan to use.
Choosing for RAG and agents
Test representative queries and inspect the evidence passed to the LLM. Useful content, response size and reliable source URLs all affect the final answer. For workflows that repeat searches or scrape pages, compare the supporting task results below.
Supporting Comparisons: Developer Search and Multiple Searches
The main ranking above measures search. These additional benchmarks measure complete agent tasks, with separate results for search-only and search with scraping.
Exa vs Tavily for Developer Search
Task completion on 100 held-out documentation questions. Each task uses the same agent and scoring method. Full developer benchmark →
Search-only
| Rank | Vendor | Endpoint & configuration | Official docs | Task completion | Median task time | Median task cost | Avg search time | Median task tokens | API list price |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Exa fast | POST /search type=fast | Exa official docs ↗ | 66.3 ± 1.5 | 20s | $0.103$0.035 search$0.068 token cost | 626ms | 22,344 | $0.007 / search |
| 2 | Exa instant | POST /search type=instant | Exa official docs ↗ | 61.3 ± 2.9 | 21s | $0.105$0.035 search$0.070 token cost | 447ms | 22,423 | $0.007 / search |
| 3 | Tavily basic | POST /search search_depth=basic | Tavily official docs ↗ | 51.0 ± 5.3 | 26s | $0.097$0.040 search$0.057 token cost | 1.72s | 16,299 | $0.008 / search |
Search and scrape
| Rank | Vendor | Endpoint & configuration | Official docs | Task completion | Median task time | Median task cost | Avg search time | Median task tokens | API list price |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Exa deep | POST /search type=deepPOST /contents | Exa official docs ↗ | 83.0 ± 1.0 | 37s | $0.127$0.057 search+scrape$0.070 token cost | 3.97s | 23,660 | $0.012 / search$0.001 / scrape |
| 2 | Exa auto | POST /search type=autoPOST /contents | Exa official docs ↗ | 81.7 ± 1.1 | 23s | $0.107$0.032 search+scrape$0.074 token cost | 1.19s | 27,433 | $0.007 / search$0.001 / scrape |
| 3 | Tavily advanced | POST /search search_depth=advancedPOST /extract extract_depth=advanced | Tavily official docs ↗ | 60.0 ± 2.0 | 46s | $0.153$0.085 search+scrape$0.069 token cost | 3.41s | 26,269 | $0.016 / search$0.0032 / scrape |
| 4 | Tavily basic | POST /search search_depth=basicPOST /extract extract_depth=basic | Tavily official docs ↗ | 59.0 ± 1.7 | 32s | $0.113$0.043 search+scrape$0.070 token cost | 1.50s | 27,405 | $0.008 / search$0.0016 / scrape |
Exa vs Tavily for Multiple Searches
F1 measures how well the agent finds the complete answer across 45 questions. Task cost includes LLM and search/scrape API spend. Full multi search benchmark →
Search-only
| Provider | Endpoint & configuration | Official docs | F1 | Precision | Recall | Exact set | Median time | Mean turns | Median task cost | API list price |
|---|---|---|---|---|---|---|---|---|---|---|
| Exa deep | POST /search type=deepexa-deep | Exa official docs ↗ | 45.4 ± 2.0 | 83.2 ± 3.0 | 33.7 ± 1.4 | 1.5 ± 1.3 | 89.5 s | 7.4 | $0.717$0.156 search$0.556 token cost | $0.012 / search |
| Exa instant | POST /search type=instantexa-instant | Exa official docs ↗ | 43.3 ± 1.0 | 82.6 ± 3.8 | 32.2 ± 0.5 | 3.0 ± 1.3 | 49.8 s | 7.3 | $0.652$0.091 search$0.559 token cost | $0.007 / search |
| Tavily advanced | POST /search search_depth=advancedtavily-advanced | Tavily official docs ↗ | 41.1 ± 2.3 | 83.7 ± 4.2 | 29.8 ± 1.7 | 2.2 ± 0.0 | 92.2 s | 7.4 | $1.029$0.224 search$0.808 token cost | $0.016 / search |
| Tavily basic | POST /search search_depth=basictavily-basic | Tavily official docs ↗ | 36.5 ± 3.2 | 82.5 ± 1.9 | 25.7 ± 2.7 | 1.5 ± 1.3 | 63.0 s | 7.7 | $0.615$0.112 search$0.504 token cost | $0.008 / search |
F1, precision, recall, and exact-set accuracy are percentages reported as mean ± sample SD across three independent runs; each run aggregates all 45 questions. SD is measured in percentage points. Median time is the median end-to-end time across all runs for each vendor. Median task cost is the median of LLM $ plus search API $ per agent run. API list price is the PAYG unit rate of the search endpoint the harness calls.
Search and scrape
| Provider | Endpoint & configuration | Official docs | F1 | Precision | Recall | Exact set | Median time | Mean turns | Median task cost | API list price |
|---|---|---|---|---|---|---|---|---|---|---|
| Exa deep | POST /search type=deepexa-deep | Exa official docs ↗ | 48.2 ± 2.1 | 89.4 ± 1.2 | 36.0 ± 2.2 | 2.2 ± 2.2 | 95.8 s | 7.5 | $0.683$0.156 search+scrape$0.526 token cost | $0.012 / search$0.001 / scrape |
| Exa instant | POST /search type=instantexa-instant | Exa official docs ↗ | 44.9 ± 0.9 | 85.9 ± 2.5 | 33.5 ± 0.9 | 5.2 ± 1.3 | 52.5 s | 7.5 | $0.653$0.091 search+scrape$0.557 token cost | $0.007 / search$0.001 / scrape |
| Tavily advanced | POST /search search_depth=advancedtavily-advanced | Tavily official docs ↗ | 41.0 ± 1.3 | 89.4 ± 6.0 | 29.1 ± 0.7 | 2.2 ± 0.0 | 92.6 s | 7.3 | $0.898$0.195 search+scrape$0.685 token cost | $0.016 / search$0.0032 / scrape |
| Tavily basic | POST /search search_depth=basictavily-basic | Tavily official docs ↗ | 38.1 ± 1.4 | 81.2 ± 5.9 | 27.8 ± 0.7 | 2.2 ± 0.0 | 62.0 s | 7.8 | $0.610$0.112 search+scrape$0.507 token cost | $0.008 / search$0.0032 / scrape |
F1, precision, recall, and exact-set accuracy are percentages reported as mean ± sample SD across three independent runs; each run aggregates all 45 questions. SD is measured in percentage points. Median time is the median end-to-end time across all runs for each vendor. Median task cost is the median of LLM $ plus search/scrape API $ per agent run. API list price is the PAYG unit rate of the search (and scrape) endpoint the harness calls.
How We Compare Exa and Tavily
The primary search benchmark uses 300 company-news questions, one query and up to 10 results per question. The answer model and judge stay fixed while the search API changes. Accuracy is the share of answers scored correct. Speed is mean API response time. Prices use the tested endpoint’s list rate.
The developer benchmark measures documentation tasks; the multiple-search benchmark measures company-discovery questions with several constraints. Each has its own scoring method. Public samples show the format, and the scoring questions are held out.
Search data last queried: 12 Sept 2026. Pricing checked 19 September 2026. Editorial update: 19 September 2026.
Open source search benchmark code · Public dataset sample · Full search methodology and results
Exa vs Tavily: Search API and Pricing FAQ
Did Nebius acquire Tavily, and when?
Yes. Nebius announced the agreement on 10 February 2026. Its subsequent financial filing records the acquisition of 100% of Tavily on 19 February 2026. Tavily continues under its own brand within Nebius.
What changed for Tavily API users after the Nebius acquisition?
Tavily’s announcement said the API and customer data policies would remain unchanged, with zero data retention continuing as a core commitment. Nebius’s May update described continued support for Tavily’s existing developer ecosystem. For an integration today, use the current API docs, pricing and contractual terms.
Is Exa a good Tavily alternative for AI search?
Exa is a Tavily alternative for web search in AI apps, agents and RAG. In this benchmark, Exa fast reaches 99.3% accuracy versus 93.0% for Tavily advanced. The fastest tested modes are Exa instant at 398ms and Tavily basic at 1.88s. Compare request settings, returned content and your billing plan before switching.
Which Tavily alternatives should I compare besides Exa?
Compare Brave, Parallel, Firecrawl and other providers in the full web search benchmark using the same accuracy, speed and cost metrics. Use the dedicated Tavily vs Brave comparison for that pair. This page provides the detailed Exa alternative, including tested modes, pricing per 1,000 searches and official docs.
Do I need to migrate from Tavily after the Nebius acquisition?
Tavily’s published announcement supports continuing with the existing API. Evaluate a migration when another API better fits your measured accuracy, latency, returned content or expected bill. Map Tavily’s search_depth and content settings to Exa’s type and content options, then test both on your application’s queries.
Exa vs Tavily: which web search API is better?
Exa fast reaches 99.3% search accuracy, compared with 93.0% for Tavily advanced on the same 300 questions. Compare the tested modes below for speed and price. These search results are the primary ranking; developer tasks and multiple searches have separate supporting comparisons.
Is Exa or Tavily faster?
The fastest tested Exa mode is Exa instant at 398ms mean search latency. The fastest tested Tavily mode is Tavily basic at 1.88s. These are measured search times for the configurations in the main table; application response time also includes LLM generation and other steps.
How much do Exa and Tavily cost per 1,000 searches?
For the tested modes, Exa instant and fast cost $7.00 per 1,000 queries. Tavily basic costs $8.00 and advanced $16.00 on pay-as-you-go pricing. Monthly plans and free credits can change the bill; the pricing section shows those separately.
Is Tavily cheaper than Exa on a monthly plan?
Tavily’s Growth plan provides 100,000 credits for $500 a month. Fully used, that is $5 per 1,000 basic searches or $10 per 1,000 advanced searches. Compare that commitment and your expected usage with Exa’s per-request price. Effective cost rises when a prepaid plan is underused.
Which is best for RAG and LLM agents?
Start with the search accuracy and evidence recall results, then compare how much useful text each endpoint returns and how quickly your application needs it. Match the mode to your queries and token budget. For agents that make several searches or scrape pages, use the supporting comparisons on this page.
Do Exa and Tavily return page content?
Both provide search results with source URLs and content options. Exa supports text and highlights. Tavily returns result content and can include cleaned raw content. Choose the response fields in the official docs, then test evidence quality and response size with your own questions.
What does search accuracy mean in this comparison?
It is the percentage of 300 company-news questions answered correctly from one search with up to 10 results. Each API receives the same questions, and the answer model and judge are held constant. AR@1 and AR@5 measure whether answer-bearing evidence appears in the first result or first five results. The code and public dataset sample are linked below; scoring uses held-out questions.
How are search prices different from total application cost?
Price per 1,000 queries covers search requests. Your application also pays for LLM input and output tokens, repeated searches and any additional scrape calls. The supporting agent tables report median task cost, including LLM and search/scrape API spend. Cost per 1,000 correct answers in the detailed search table divides query cost by measured accuracy.
Where are the official Exa and Tavily docs and pricing?
Each tested mode links to official search documentation and vendor pricing. The integration comparison explains the key request settings; the pricing section states the billing basis and verification date.
How do Exa and Tavily compare with Brave?
Use the linked Exa vs Brave and Tavily vs Brave comparisons, or the full web search benchmark, to compare the same metrics across more providers. This page ranks the tested Exa and Tavily search modes.