Arena benchmark
LM Arena Search Factuality Leaderboard
LM Arena Search Factuality model ratings and results.
Updated Aug 17, 2026. Data as of 2026-08-17
Models24
Categories1
Result typerating
Search Factuality Ranking
Models are ordered by their Arena rank.
Sort by
Rank | Model | Organization | Rating / score | Votes | Observations | Result date |
|---|
| Rank01 | ModelOP | Organizationopenai | Rating / score1244.1 | Votes67,534 | ObservationsN/A | Result date |
| Rank02 | ModelAN | Organizationanthropic | Rating / score1227.3 | Votes112,157 | ObservationsN/A | Result date |
| Rank03 | ModelOP | Organizationopenai | Rating / score1222.7 | Votes52,556 | ObservationsN/A | Result date |
| Rank04 | ModelAN | Organizationanthropic | Rating / score1216.5 | Votes20,630 | ObservationsN/A | Result date |
| Rank06 | ModelOP | Organizationopenai | Rating / score1212.1 | Votes87,310 | ObservationsN/A | Result date |
| Rank07 | ModelAN | Organizationanthropic | Rating / score1211.5 | Votes68,482 | ObservationsN/A | Result date |
| Rank08 | ModelOP | Organizationopenai | Rating / score1203.8 | Votes59,855 | ObservationsN/A | Result date |
| Rank09 | ModelAN | Organizationanthropic | Rating / score1201.0 | Votes111,907 | ObservationsN/A | Result date |
| Rank10 | ModelAN | Organizationanthropic | Rating / score1199.3 | Votes48,610 | ObservationsN/A | Result date |
| Rank11 | ModelGO | Organizationgoogle | Rating / score1199.0 | Votes89,924 | ObservationsN/A | Result date |
| Rank12 | ModelXA | Organizationxai | Rating / score1199.0 | Votes87,067 | ObservationsN/A | Result date |
| Rank15 | ModelGO | Organizationgoogle | Rating / score1193.7 | Votes37,134 | ObservationsN/A | Result date |
| Rank16 | ModelAN | Organizationanthropic | Rating / score1186.9 | Votes17,604 | ObservationsN/A | Result date |
| Rank17 | ModelGO | Organizationgoogle | Rating / score1180.2 | Votes125,762 | ObservationsN/A | Result date |
| Rank18 | ModelAN | Organizationanthropic | Rating / score1179.7 | Votes61,439 | ObservationsN/A | Result date |
| Rank19 | ModelXA | Organizationxai | Rating / score1173.3 | Votes68,373 | ObservationsN/A | Result date |
| Rank20 | ModelAN | Organizationanthropic | Rating / score1161.3 | Votes105,575 | ObservationsN/A | Result date |
| Rank21 | ModelGO | Organizationgoogle | Rating / score1160.7 | Votes66,208 | ObservationsN/A | Result date |
| Rank22 | ModelAN | Organizationanthropic | Rating / score1153.6 | Votes60,046 | ObservationsN/A | Result date |
| Rank25 | ModelXA | Organizationxai | Rating / score1147.0 | Votes30,241 | ObservationsN/A | Result date |
| Rank26 | ModelOP | Organizationopenai | Rating / score1143.3 | Votes3,921 | ObservationsN/A | Result date |
| Rank27 | ModelAN | Organizationanthropic | Rating / score1138.1 | Votes14,073 | ObservationsN/A | Result date |
| Rank28 | ModelXA | Organizationxai | Rating / score1131.3 | Votes3,149 | ObservationsN/A | Result date |
| Rank30 | ModelOP | Organizationopenai | Rating / score1128.1 | Votes3,706 | ObservationsN/A | Result date |
Search Factuality Rating Distribution
A closer view of the leading model ratings.
The Top AI Models for Search Factuality
The first five models in this source ranking, with price and output speed included when a canonical model match is available.
Ranking basisThis search factuality AI model leaderboard uses the source rating and Arena rank. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01OPGPT-5.5OpenAI
- rating
- 1,244
- Price
- $5.0 input / $30 output per 1M tokens
Strengths
- Ranks #1 on Search Factuality
- 67,534 votes support the result
- Matched official PAYG token pricing is available
Considerations
- Arena results are independent from the LLMBoard capability score
- 02ANClaude Opus 4.6Anthropic
- rating
- 1,227
- Price
- $5.0 input / $25 output per 1M tokens
- Speed
- Up to 17 tok/s via Anthropic
Strengths
- Ranks #2 on Search Factuality
- 112,157 votes support the result
- Matched official PAYG token pricing is available
Considerations
- Arena results are independent from the LLMBoard capability score
- 03OPGPT-5.2OpenAI
- rating
- 1,223
- Price
- $1.8 input / $14 output per 1M tokens
- Speed
- Up to 100 tok/s via OpenAI
Strengths
- Ranks #3 on Search Factuality
- 52,556 votes support the result
- Matched official PAYG token pricing is available
Considerations
- Arena results are independent from the LLMBoard capability score
- 04ANClaude Fable 5Anthropic
- rating
- 1,216
- Price
- $10 input / $50 output per 1M tokens
- Speed
- Up to 9.1 tok/s via Anthropic
Strengths
- Ranks #4 on Search Factuality
- 20,630 votes support the result
- Matched official PAYG token pricing is available
Considerations
- Arena results are independent from the LLMBoard capability score
- 06OPGPT-5.4OpenAI
- rating
- 1,212
- Price
- $2.5 input / $15 output per 1M tokens
- Speed
- Up to 9.1 tok/s via OpenAI
Strengths
- Ranks #6 on Search Factuality
- 87,310 votes support the result
- Matched official PAYG token pricing is available
Considerations
- Arena results are independent from the LLMBoard capability score
Selection summary
Best AI Models for Search Factuality
GPT-5.5 currently leads Search Factuality at 1,244. The best AI model for this use case may change when price, speed and independent benchmark capability are considered.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.
What is the Search Factuality Arena?
What Search Factuality measures and how to read its results.
Search Factuality reports rating values and currently defines 1 evaluation category.
Arena results are preference or task-outcome signals. They are not automatically equivalent to capability benchmark scores.
FAQ
Common questions about Search Factuality.
Which model leads Search Factuality?
GPT-5.5 is currently ranked first.
What does Search Factuality measure?
This Arena reports rating values across 1 configured evaluation category.
Does this result affect the LLMBoard score?
No. Arena results remain a separate preference or task-outcome signal.
How many models are included?
24 model results are currently shown.