AI benchmarks.
12+ months of history.
Explore 649 benchmark pages and 784 model profiles. Query current rankings and eligible historical records through REST or MCP.
Arena overall snapshots date back to August 2024. Coverage varies by source, model, and date.
Check source and date coverage free before subscribing.
Catalog totals describe the website. API results and historical coverage vary by source.
A current ranking helps you choose today. Historical records let you inspect how the published results changed.
Follow Arena overall ratings, ranks, and confidence intervals across retained snapshots. Source-reported benchmark runs date back to January 2025. Research includes eligible records beyond 12 months.
Check the dates availableInspect the exact source model, original units, and reported dates. Saved BenchAlign rankings retain their scoring version, so a methodology change stays visible in your analysis.
Read the scoring methodologyAsk an MCP client how a model's rank changed, or query REST from a notebook or application. One API key works with both. Resolve model IDs and check coverage before requesting history.
See how to connectAll plans use the same API key for REST and MCP.
Current data
$0 USD/month
Look up models and benchmarks. Read current BenchAlign rankings and licensed benchmark results.
Create a free accountNo payment details required. Historical queries require Pro or Research.
Most popular
Current + entire available archive
$249 USD/month
Investigate changes beyond the last three months. Get the entire eligible archive and 5× the monthly reads of Pro.
Get Research accessCoverage varies by source, model, and date. Check it free in your account before subscribing. Checkout opens when available.
Current + three months of history
$49 USD/month
Query retained rankings, publisher snapshots, and dated runs from the last three rolling calendar months.
Get Data ProNeed records older than three months? Choose Research. Check coverage before subscribing.
Access plans cover queries and internal analysis. Commercial display and redistribution need a separate license. See usage terms.
Radar + Data includes Radar and Pro’s three-month history window.
$59 USD/month
See the bundleRadar and Data remain separate products, with separate API keys and allowances. Existing monthly, annual, and legacy Radar subscriptions keep their terms until period end before bundle checkout opens.
Use an MCP client for natural-language questions, or call the REST API from your own application.
All plans
Read final BenchAlign scores and ranks for a published surface. Search the catalogs and filter benchmarks by capability, including coding, reasoning, or math.
benchlm_list_current_rankingsPro · Research
Read recorded scores and ranks over time, with the saved methodology version. Retrieve a ranking snapshot to compare the models recorded at a date.
benchlm_rank_historyPro · Research
Find catalog additions, removals, and reintroductions in the archive. Dates describe saved repository states, rather than a publisher’s release date.
benchlm_benchmark_events| Query | Fields and coverage | Plan | MCP tool |
|---|---|---|---|
| Account usage | Current plan, successful and remaining monthly reads, request-rate usage, and both reset times. Usage checks do not consume a monthly read. | Free | benchlm_get_usage |
| Current BenchAlign rankings | Final scores, published ranks, evidence labels, interval bounds, scoring date, and methodology version for each available surface. | Free | benchlm_list_current_rankings |
| Current model rankings | One model’s final score and rank across every available BenchAlign surface, with explicit gaps where it is unranked. | Free | benchlm_get_current_model_rankings |
| Model catalog | Names, creators, canonical IDs, source links, and sourced release dates with their date precision. Search by name or filter by creator. | Free | benchlm_list_models / benchlm_get_model |
| Benchmark catalog | Names, catalog IDs, source references, and curator categories and tags. Search by name or filter by category, tag, and relationship; discovery labels do not establish result coverage. | Free | benchlm_list_benchmarks |
| Current benchmark results | All admitted current source results, with native units and exact profiles. Examples include Epoch AI, BullshitBench v2, and Arena overall. Search by category or model name. | Free | benchlm_list_current_benchmark_results |
| Benchmark result coverage | Available source boards, row counts, metrics, units, and source licenses. Coverage checks do not use a read. | Free | benchlm_benchmark_result_coverage |
| Published evaluation runs | Source-reported benchmark runs for an exact board and model profile. Run dates are distinct from archive commit dates; undated records are identified. | Pro · Research | benchlm_benchmark_result_history |
| Published benchmark snapshots | Follow an exact source model’s published rating, rank, and confidence interval over time. Arena overall history preserves three scoring methods and publisher dates. | Pro · Research | benchlm_benchmark_result_snapshot_history |
| Historical coverage | Available methods, retained date windows, and delivery coverage. Check this before requesting history. | Free | benchlm_history_coverage |
| BenchAlign history | Saved BenchLM scores, ranks, evidence labels, ranking surfaces, and methodology versions for an exact model. | Pro · Research | benchlm_rank_history |
| Ranking at a date | Stored BenchAlign ranking rows as of a UTC date. The response identifies incomplete coverage and preserves the saved scoring version. | Pro · Research | benchlm_ranking_snapshot |
| Benchmark history | Cleared measurements for an exact model, benchmark, and protocol. Availability differs by source; this is not every score on the website. | Pro · Research | benchlm_benchmark_history |
| Benchmark changes | First recorded catalog definitions and results, additions, removals, and reintroductions where retained. | Pro · Research | benchlm_benchmark_events |
Current ranking responses contain final BenchAlign outputs. Benchmark result queries contain licensed source scores, including GPQA Diamond, SWE-bench Verified, and FrontierMath evaluated by Epoch AI. BullshitBench v2 includes judge ratings on a 0–2 scale, refusal rates, and published ranks from one pinned snapshot. Source-reported runs retain their original units and dates; artifact publication timestamps do not establish run dates. Arena overall snapshot history contains the publisher’s ratings, ranks, and confidence intervals across retained dates, with scoring methods kept separate. Benchmark measurements and evaluation runs do not establish a leaderboard rank. Missing records stay missing.
Sign in by email. Free starts without a payment method.
Keep it in your application’s or MCP client’s secret settings.
Use REST, or connect an MCP client with Streamable HTTP.
curl 'https://data.benchlm.ai/v1/rankings/current?surface=coding&limit=10' \
-H 'Authorization: Bearer YOUR_DATA_KEY'https://data.benchlm.ai/mcpSend Authorization: Bearer YOUR_DATA_KEY. Supported MCP protocol versions: 2026-07-28, 2025-11-25, and 2025-06-18.
curl 'https://data.benchlm.ai/mcp' \
-H 'Authorization: Bearer YOUR_DATA_KEY' \
-H 'Content-Type: application/json' \
-H 'Accept: application/json, text/event-stream' \
-H 'MCP-Protocol-Version: 2025-11-25' \
--data '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"benchlm_list_current_rankings","arguments":{"surface":"coding","limit":10}}}'Check the records you need before you choose a plan. Coverage and usage checks are free, including when your monthly reads are exhausted.
benchlm_get_usageNo. Retained Arena overall snapshots date back to August 2024, and source-reported benchmark runs reach January 2025. Other sources and saved BenchAlign rankings begin later. Research removes the history age cap; it does not fill missing records. Check the dates available for your exact source and model.
Research includes all retained eligible history and 500,000 successful reads per month for $249 USD. Data Pro includes three rolling calendar months and 100,000 reads for $49 USD. Both allow 60 requests per minute. Choose Research when older records or a larger monthly allowance matter to your work.
Yes. A free account gives you current rankings, licensed source results, model and benchmark catalogs, and coverage checks. The coverage tools identify available sources and date windows without using your monthly reads. Check your model and benchmark first, then choose the plan that includes the history you need.
One successful data page uses one read. Initialization, discovery, coverage and usage checks, and unsuccessful queries do not use the monthly allowance. All keys on an account share usage across REST and MCP.
Free renews on your account’s activation anniversary each month. Pro and Research follow your subscription period. Short months use their last day. Your account and usage responses show the exact monthly and request-rate reset times.
The API and MCP identify the monthly read allowance or request-rate limit, with a reset time and retry delay. Free accounts get a Data Pro upgrade link. Eligible Pro accounts can upgrade to Research with a prorated price difference and a fresh 500,000-read allowance for the remaining subscription period. You can receive usage emails at 75%, 90%, and 100%; manage them in your account. Usage and coverage checks stay available.
Pro covers three rolling calendar months by UTC date; Research has no age cap on retained eligible records. Coverage varies by source. Historical pages contain up to 500 rows. Saved-state history uses 90-day windows and cursors. Source-run and publisher snapshot history take an exact board and source model ID, with date filters of up to 90 days. Snapshot history supports cursors and preserves publisher dates; run history preserves reported evaluation times. Check the relevant coverage tool first.
Free, Pro, and Research are access plans for queries and internal analysis. Paid plans add reads and available history; they do not include a commercial display or redistribution license.
Publishing BenchLM-owned outputs or the compiled feed in a customer-facing commercial website, app, or product needs a separately agreed display license. Redistributing that material through an API, downloads, or resale needs separately agreed scope. Request a commercial license.
Individual quotations and attributed embeds keep their existing permissions. Independently licensed publisher measurements keep their own commercial-use and redistribution rights. Retain attribution, source notices, and licenses. Read the Data access terms.
Research gives you all retained eligible history for $249 USD/month. Check your sources and dates with a free account first.
Looking for the website’s existing downloads and their terms? Open the Data hub and downloads.