Frontier models

Model
explorer.

Every language model Artificial Analysis benchmarks, in one place. Set your priorities, then filter, sort, and see where each model sits on intelligence, speed, and price.

699 models59 labsUpdated 14 Oct 2026

Your priorities

Find the model for the job.

Weight what matters. The top ten, the map, and the grid update as you go, and the address bar keeps your settings, so you can share the view.

How much does each matter?

Very important
Low
Low

What kind of intelligence?

Pick one or more. AA averages available focus scores and falls back to general intelligence if none is available. Science is our additional focus, using the same scoring method.

Benchmarks behind each focus

As audited on 5 October 2026: General: Artificial Analysis Intelligence Index; Coding: AA coding focus: Terminal-Bench 4.0; Agentic: AA agentic focus: Briefcase, GDPval, AutomationBench, Terminal-Bench; Reasoning: AA math and reasoning focus: CritPt and HLE; Science: Our extension: available GPQA Diamond and SciCode scores; Instruction following: AA instruction-following focus: IFBench; Long context: AA long-context focus: AA-LCR; Knowledge: AA knowledge focus: AA-Omniscience; Document creation: AA document-creation focus: GDPval-AA; Low hallucination: AA low-hallucination focus: AA-Omniscience. A g after a focus score means the model has no score for that focus, so the general Intelligence Index is shown, as AA’s method does.

Models and settings
Select models

Select a family or individual settings. Other filters still apply.

Save a selection on this browser, or share the current URL.

Labs
All labs

Thresholds keep models strictly above the values entered. Intelligence always uses the general index; speed uses output tokens per second. Leave a field blank to disable that filter. Models without that measurement are excluded when its filter is active.

AA requirements and providers

These requirements use AA's recommender cohort. Size applies when open weights is selected; AA retains models with unknown parameter counts.

API providers

Top ten

Ranked for your priorities.

The match score follows Artificial Analysis’s recommender method. It scores AA’s eligible cohort from 0 to 100; models outside the cohort stay in the grid, without a match score. Filters narrow the shortlist without changing the scoring scale. How it works.

Artificial Analysis
  1. Turn on JavaScript to rank models for your priorities. The grid below lists the highest-scoring models without it.

Source: Artificial Analysis (artificialanalysis.ai). Scoring method: AA Model Recommender, implemented by Promise & Proof.

Discuss this shortlist →

The map

Where every model in view sits.

Your top ten in orange. The frontier (the orange line) is the most intelligence available at each price, or at each speed. Hover, or tab to the chart and use the arrow keys.

Artificial Analysis

The map needs JavaScript; the grid below has every value.

Task cost is AA’s weighted average USD per Intelligence Index task. Task time is weighted decode time in minutes, excluding time to first token and overhead. Token price is the 3:1 input/output blend per million tokens; output speed is tokens per second. Missing measurements are omitted from the map. Source: Artificial Analysis (artificialanalysis.ai).

The grid

Every language model, sortable.

Sort by a column heading, or use the sort menu on a phone. Benchmark rates are percentages; Elo and index columns retain their named scales. Task time is in minutes. The grid includes current and legacy benchmarks supplied by the sources. A dot means Artificial Analysis has not published that measure.

Source: Artificial Analysis (artificialanalysis.ai)
#
1Claude Opus 5.5 (Max, Default Fallback)Anthropic·57.657.659.6··59.0··66.961.4·31.784.7···46.466.258.61866.368.369.526.238.2·87.756.24.2·66.71805.061.22145.71717.31915.2···60.763.763.360.560.465.628.758.326.90.835.99.20.098.59.88.14.311.752.2$5.98213.15$870895673.6$8.00$4.00$20Sep 2026
2Claude Sonnet 5.5 (Max, Default Fallback)Anthropic·56.056.063.6··53.3··61.055.0·31.482.7···32.353.947.01837.866.971.825.8···57.52.8·75.01820.465.22119.01720.71863.1···57.259.656.458.158.460.8············45.8$5.46115.31$7260135430.2$4.00$2.00$10Sep 2026
3Claude Opus 5.5 (Xhigh, Default Fallback)Anthropic·56.056.059.6··61.9··65.057.5·31.784.7···42.665.465.71835.866.865.026.6··86.6····1765.860.62100.11642.11116.2···58.461.660.8·58.763.0·············$3.4598.43$405782163.7$8.00$4.00$20Sep 2026
4Muse Wonder Full (Max)Meta·55.855.857.1·····62.653.6·32.983.7···30.143.223.01857.167.971.332.6····13.1··1729.361.32116.31524.0····54.257.854.7·57.060.7·····················Oct 2026
5Claude Opus 5.5 (High, Default Fallback)Anthropic·53.653.656.6··49.0··60.455.6·30.982.7···40.664.667.61702.560.163.228.8··85.8····1687.958.91985.81552.9570.6···55.959.259.0·56.260.6·············$1.8234.53$21728360.7$8.00$4.00$20Sep 2026
6Claude Fable 5.1 (Max, Default Fallback)Anthropic·53.453.452.091.4·43.3·47.263.159.193.729.785.3···43.567.272.61754.162.759.426.249.5··57.56.4·71.11673.561.51968.81492.12066.9·81.6·56.459.760.558.056.562.725.050.823.50.838.416.00.099.28.18.95.77.649.0$7.63011.41$1312971258.5$20$10$50Sep 2026
7Claude Fable 5.1 (Xhigh, Default Fallback)Anthropic·53.253.255.191.0···45.860.958.793.431.183.0···42.466.270.51732.861.657.826.2·······1655.760.61929.51478.91619.3·80.7·56.258.460.8·56.862.4·············$5.9789.91$906363122.5$20$10$50Sep 2026
8GPT-6 Astra (Max)OpenAI·52.752.759.188.4·63.3·41.456.554.796.131.780.7···43.462.651.31572.853.668.531.048.6·86.951.28.6·35.01570.152.01760.51549.5864.7·76.9·55.257.358.551.754.759.833.463.336.90.037.813.30.0100.05.05.14.54.947.0$3.2587.84$532457284.2$20$10$50Sep 2026
9Gemini 4 Argon (High)Google·52.652.657.1·····61.857.1·27.179.7···42.449.915.11624.056.277.521.8···48.8···1487.864.61564.81304.2530.3···53.856.260.0·55.260.4············42.6$1.990·$2407··$4.00$2.00$10Sep 2026
10Muse Wonder (Max)Meta·52.452.452.0·····60.350.1·28.380.7···25.541.126.51770.363.568.130.6····10.0··1658.761.21962.41475.81133.0···51.354.852.0·53.857.7·············$4.5166.83$573313139.2$8.00$4.00$20Oct 2026
11GPT-6 Astra (Xhigh)OpenAI·52.452.459.689.1···43.155.754.696.331.480.0···43.461.948.31554.552.767.232.2··86.2····1546.452.31710.41518.3597.3·75.9·54.456.758.2·54.759.5·············$2.3095.28$380353166.2$20$10$50Sep 2026
12Claude Sonnet 5.5 (Xhigh, Default Fallback)Anthropic·51.951.957.1··52.4··57.350.0·31.179.7···23.553.062.91729.561.565.524.6·······1751.862.22031.81646.0634.8···52.356.252.2·53.955.9·············$2.0127.39$217710534.5$4.00$2.00$10Sep 2026
13GPT-6.1 Sol (Max)OpenAI·51.851.856.1··58.1··54.252.9·31.783.0···41.562.154.31591.654.664.931.0··86.050.06.9·33.91561.055.71780.31450.8210.4···54.057.057.751.253.959.3············49.1$0.72411.59$108255271.7$4.00$2.00$10Sep 2026
14Claude Opus 5.5 (Medium, Default Fallback)Anthropic·51.251.252.5··43.3··59.354.7·27.784.3···40.364.568.41583.754.261.225.6··85.7····1627.457.01878.11502.5400.2···54.157.557.4·53.759.7·············$1.3363.38$16277926.6$8.00$4.00$20Sep 2026
15Claude Fable 5.1 (High, Default Fallback)Anthropic·51.251.252.089.9···43.158.755.990.630.383.7···40.864.968.81633.656.755.326.8·······1580.857.21837.01409.71037.3·79.1·53.856.158.1·54.560.2·············$3.9136.82$52425818.0$20$10$50Sep 2026
16GPT-6.1 Sol (Xhigh)OpenAI·51.051.054.0·····55.752.6·31.779.7···40.960.850.91538.751.966.631.8··85.1····1504.051.61699.41426.994.5···53.256.056.8·53.258.3·············$0.3935.09$66257186.0$4.00$2.00$10Sep 2026
17GPT-6 Astra (High)OpenAI·50.950.954.089.9···40.055.453.194.928.980.0···43.761.144.81523.951.266.631.0··86.4····1505.950.81678.21462.8433.5·77.1·53.255.557.9·53.058.4·············$1.7253.77$29255295.1$20$10$50Sep 2026
18Claude Opus 5 (Max)Anthropic · Retired by AA·50.850.849.089.1·28.6·42.156.454.993.229.179.3···37.160.960.81721.961.156.621.6··84.753.8·47.555.61660.157.21916.51567.81619.2·78.0·54.956.657.353.354.060.8·············$5.85812.95$72755699.5$10$5.00$25Jul 2026
19GPT-6.1 Sol (High)OpenAI·50.250.251.5·····55.851.4·30.082.3···41.560.849.41509.550.564.532.0··84.9····1464.549.11650.81375.576.2···52.654.956.2·51.957.6·············$0.3193.84$5215656.7$4.00$2.00$10Sep 2026
20Claude Opus 5 (Xhigh)Anthropic · Retired by AA·49.749.746.588.0···43.355.754.493.727.780.3···35.459.559.51691.559.653.221.0··84.0···58.31634.858.01908.31493.01297.8·77.0·54.354.356.052.952.860.2·············$4.87810.98$58685742.7$10$5.00$25Jul 2026
21Claude Fable 5 (Max, Opus 4.8 Fallback)Anthropic · Retired by AA·49.649.642.484.662.9·98.538.161.055.592.628.682.363.5··43.365.363.61609.555.554.124.0···48.8·51.164.41539.656.01663.71474.82029.6·76.5·53.855.558.053.853.060.4·············$8.74610.63$1116165117.3$20$10$50Jun 2026
22GPT-6 Astra (Medium)OpenAI·49.649.649.589.5···35.554.252.793.929.179.7···42.260.646.51485.449.364.630.4··85.1····1460.647.51645.41414.0358.5·76.7·52.254.256.6·51.757.8·············$1.5413.13$2434527.2$20$10$50Sep 2026
23Claude Fable 5.1 (Medium, Default Fallback)Anthropic·48.948.944.988.0···41.056.453.888.629.184.7···37.663.169.11545.452.354.726.8·······1528.456.11771.61341.4780.8·77.1·51.353.956.7·51.557.8·············$2.9834.97$39835912.6$20$10$50Sep 2026
24Claude Opus 5 (High)Anthropic · Retired by AA·48.148.146.087.6···44.755.452.893.728.379.0···33.758.961.21594.554.753.619.6··82.4···59.41561.353.81813.71448.7947.2·76.5·51.453.254.552.051.658.2·············$3.6138.66$43325425.7$10$5.00$25Jul 2026
25Muse Spark 1.3 (Max)Meta·48.148.133.384.3·11.0·50.558.848.793.524.983.0···25.043.632.91681.059.057.926.633.2··37.58.9·43.31581.358.71699.31525.4369.4·75.8·49.851.250.949.749.456.3············40.6$1.6057.81$200012841.8$2.00$1.25$4.25Sep 2026
26GPT-6.1 Sol (Medium)OpenAI·47.847.848.0·····53.249.9·27.783.3···40.060.451.61448.847.462.630.0··83.9····1360.544.01539.71259.042.3···50.452.654.4·49.756.0·············$0.2142.67$361507.5$4.00$2.00$10Sep 2026
27GPT-6 Sol (Max)OpenAI · Retired by AA·47.647.643.9··30.0··57.647.9·30.983.7···27.154.560.11508.250.461.625.249.4·82.9·3.6·16.11480.946.31618.51516.0238.8···48.652.551.143.648.654.436.864.246.30.036.49.20.0100.02.01.91.82.0·$1.0454.76$1536118152.5$4.00$2.00$10Sep 2026
28GPT-5.6 Sol (Max)OpenAI · Retired by AA·47.047.039.988.065.922.485.144.357.149.594.132.384.072.7··22.059.492.21608.555.460.127.256.2·83.447.5·42.926.11480.441.81538.11654.8365.8·77.4·49.054.349.845.448.952.9·············$1.9885.60$346590138.4$8.00$4.00$20Jul 2026
29Claude Fable 5.1 (Low, Default Fallback)Anthropic·46.846.840.485.0···39.056.748.988.127.782.3···34.160.265.61468.348.452.228.0·······1482.555.41713.71283.1611.6·75.2·48.751.253.5·48.755.1·············$2.3713.85$3158592.6$20$10$50Sep 2026
30Claude Sonnet 5.5 (High, Default Fallback)Anthropic·46.846.843.9··34.3··53.745.8·24.678.0···20.952.064.61549.852.559.425.2·······1638.260.21884.01492.8297.6···47.151.648.7·48.351.8·············$0.8843.96$10279614.6$4.00$2.00$10Sep 2026
31Grok 4.7 (Xhigh)SpaceXAI·46.446.425.8··4.8··57.443.1·17.776.7···32.047.429.31711.160.665.620.042.1··52.59.4·15.01643.455.41947.91504.1846.6···51.954.454.147.247.455.756.468.326.974.00.00.00.00.011.78.16.115.433.8$3.73815.85$49678298.3$3.00$2.00$6.00Sep 2026
32Grok 4.7 (High)SpaceXAI·46.346.324.7·····57.842.3·18.077.0···30.947.832.41707.560.463.523.2·······1633.555.01928.21507.2674.8···51.754.454.2·46.655.3·············$2.72614.46$38817754.5$3.00$2.00$6.00Sep 2026
33MiMo-V2.6-ProXiaomi·46.346.334.8··5.7··60.949.4·26.686.3···8.434.840.61684.359.258.619.2···37.5··18.31516.054.71801.31296.234.6···46.147.844.841.846.953.256.163.326.278.60.00.00.00.00.20.20.10.245.4$0.13322.61$207464.1$0.54$0.43$0.87Sep 2026
34GPT-6 Astra (Low)OpenAI·45.845.841.988.0···32.054.149.293.126.380.0···40.559.546.91389.044.459.130.4··84.6····1257.040.21403.11155.7131.3·75.7·48.649.952.9·47.754.7·············$0.8181.49$1537502.7$20$10$50Sep 2026
35Qwen3.8 Max (0902)Alibaba·45.445.438.988.8·11.9·47.852.143.192.817.780.3···12.031.728.81665.658.356.222.840.342.482.845.0·47.620.01614.558.01955.91385.61781.5·76.2·45.647.346.141.446.452.3············43.2$5.40932.93$4935423.0$3.00$2.00$6.00Sep 2026
36Muse Spark 1.3 (Xhigh)Meta·45.145.116.785.4···47.259.747.594.126.083.0···23.141.531.51629.056.556.824.2··82.0····1486.351.61572.01488.5311.3·76.5·47.849.148.8·45.654.944.360.812.559.50.00.00.00.03.12.00.54.3·$1.3683.61$165522929.4$2.00$1.25$4.25Sep 2026
37Claude Opus 5 (Medium)Anthropic · Retired by AA·44.844.834.386.1···38.651.551.391.926.982.0···31.057.160.71490.349.554.320.0··81.6···56.11435.348.51636.31316.2477.7·74.3·49.151.151.749.747.455.9·············$2.1895.23$27325512.1$10$5.00$25Jul 2026
38GLM-5.3 (Max)Z AI·44.844.841.983.9·9.5·50.359.042.391.719.179.7···14.333.929.61650.357.562.211.246.1··38.80.336.448.31508.451.31772.01353.3514.0·74.8·44.747.642.046.747.151.236.455.824.329.00.00.00.00.02.43.61.61.440.1$2.00613.81$2503613.5$2.15$1.40$4.40Aug 2026
39Grok 4.6 (High)SpaceXAI · Retired by AA·44.344.321.288.4·6.2·50.756.542.994.917.180.3···30.548.234.31618.555.966.717.0···41.2·48.312.21534.752.81659.81520.8360.2·76.8·49.352.551.945.245.053.9·············$1.48510.73$19635640.1$3.00$2.00$6.00Aug 2026
40GPT-6 Sol (Xhigh)OpenAI · Retired by AA·44.244.230.3··17.6··55.146.3·28.081.3···26.753.858.91455.247.861.724.6··82.5····1357.241.71466.61371.8102.9···46.550.949.5·45.152.5·············$0.5212.36$85511267.4$4.00$2.00$10Sep 2026
41Grok 4.6 (Xhigh)SpaceXAI · Retired by AA·44.244.217.288.0···43.353.044.193.519.781.0···29.343.024.01645.857.367.017.2·······1544.953.81665.11532.2444.0·75.9·48.851.151.5·45.054.6·············$1.75210.00$22426251.3$3.00$2.00$6.00Aug 2026
42GPT-5.6 Sol (Xhigh)OpenAI · Retired by AA·44.044.024.789.561.4·84.838.157.147.393.128.682.371.0··21.058.891.91568.953.455.327.6··82.7···18.31437.641.31470.81597.3280.3·78.3·47.152.248.942.744.651.4·············$1.1843.82$20828467.2$8.00$4.00$20Jul 2026
43Step 5 PreviewStepFun·43.743.733.3··2.4··58.946.5·20.988.3···16.441.543.01586.654.351.014.855.638.076.435.0·47.216.71425.347.81544.21393.3215.9···45.447.146.240.845.052.847.752.515.874.80.00.00.00.01.81.60.52.138.4$1.02610.81$1190912.8$1.43$1.00$2.70Sep 2026
44Kimi K3 (Max)Kimi·43.643.612.685.0···46.059.546.993.523.488.7···19.747.653.21533.151.758.322.047.741.380.538.85.345.338.31501.051.01663.31451.4612.9·76.2·46.950.548.645.043.353.441.353.312.558.00.00.00.00.02.41.70.43.343.6$2.00016.01$3658453.2$6.00$3.00$15Jul 2026
45Claude Haiku 5.5 (Max)Anthropic·43.443.432.8··20.0··55.044.4·18.982.7···10.736.440.41620.056.035.420.8···53.80.6·63.31578.052.31905.91429.661.4···43.940.442.344.144.951.2············41.8$0.2137.05$330237332.2$0.20$0.10$0.50Oct 2026
46Grok 4.6 (Medium)SpaceXAI · Retired by AA·42.842.813.184.3···44.355.942.193.517.781.0···28.041.924.01613.255.763.217.8·······1483.652.51579.71458.6250.4·74.4·46.548.949.4·43.353.1·············$1.1448.38$15575540.9$3.00$2.00$6.00Aug 2026
47GPT-6 Sol (High)OpenAI · Retired by AA·42.442.426.3··17.1··54.944.1·25.483.7···26.853.758.11394.344.760.124.4··82.0····1267.137.91363.01277.358.2···45.749.548.2·43.151.5·············$0.3731.68$60510024.9$4.00$2.00$10Sep 2026
48GPT-5.6 Sol (High)OpenAI · Retired by AA·42.342.320.787.362.1·83.336.757.846.092.825.781.769.2··20.458.491.21503.950.255.327.8··81.8···17.21369.938.61377.51541.3187.8·77.2·45.850.647.341.843.049.8·············$0.8082.58$14878630.4$8.00$4.00$20Jul 2026
49Claude Opus 5.5 (Low, Default Fallback)Anthropic·42.342.331.3··24.3··58.648.3·17.780.7···38.963.567.61233.736.752.925.6··84.7····1279.342.51440.41147.3104.9···45.948.551.6·44.652.5·············$0.5511.35$8608111.8$8.00$4.00$20Sep 2026
50Grok 4.7 (Low)SpaceXAI·42.242.216.2·····54.939.2·13.480.0···29.649.739.91588.454.456.721.4··78.2····1499.249.31735.91405.7228.3···48.050.850.7·42.552.8·············$1.2487.55$1631636.9$3.00$2.00$6.00Sep 2026

* Harvey LAB-AA was re-scored by Artificial Analysis on 8 Oct 2026, with a new method: scores from before then are not comparable.

How the match score works

  • We use AA’s published recommender cohort and focus scores. The grid lists the language models and reasoning settings in AA’s website catalog; a model outside that cohort has no match score.
  • Intelligence is scaled linearly from the cohort minimum to maximum. Output speed and total Intelligence Index cost use log(1 + value); cheaper cost scores higher. Each scale runs from 0 to 100. AA assigns 50 when a range is constant, and 50 for missing index cost.
  • Sliders run from 0 to 5. AA doubles the intelligence weight, then takes the weighted average of intelligence, speed, and cost. All-zero priorities return no shortlist. Equal scores retain AA’s source order.
  • AA averages available selected focus scores, falling back to general intelligence when all are missing (marked g). Our Science option adds GPQA and SciCode to that same method. AA does not offer a Science option. For a model outside the cohort, Focus uses the matching benchmark where the catalog has one.
  • Models AA marks as deprecated carry a Retired by AA label and can be hidden. A shared link or saved selection that names a model, lab or provider AA no longer lists says so, and renamed models still resolve.
  • Requirements, providers, model selections, general intelligence and output speed thresholds, labs, release window, search, and the optional best-setting view filter after normalization. Your score stays stable as those filters change. Task cost and task time are separate measurements; AA’s recommender uses total index cost and output tokens per second.

AA website snapshot: . 699 models and settings; 186 with cost per task; 183 with time per task; 157 in AA's recommender cohort; 430 marked retired (deprecated) by AA. A dot means the source has not published that measurement. * marks an AA estimated Intelligence Index.

Language-model measurements and recommender inputs come from AA’s website; we refresh twice a day and retain the last successful snapshot if a source fails validation. The scoring logic was audited on 5 October 2026. We then test the shortlist on your own cases. Charts, rankings, and scores marked as ours are by Promise & Proof, based on data from Artificial Analysis; they do not represent the views or endorsement of Artificial Analysis.

Next

Tell us the job. We will test the shortlist on your cases, and prove which model fits.

Get started →hello@promiseandproof.com →