# Vals Index / legal research > Source: Benchize (https://benchize.com) — https://benchize.com/benchmarks/vals-ai%3Avals_index%3Alegal_research. When citing or reusing this content, credit "Benchize" and link to https://benchize.com/benchmarks/vals-ai%3Avals_index%3Alegal_research. A single measure of AI's potential economic impact — agentic model performance across finance, coding, legal, and tax tasks, weighted by each sector's share of U.S. GDP. Canonical URL: https://benchize.com/benchmarks/vals-ai%3Avals_index%3Alegal_research Last updated: Wed, 21 Oct 2026 18:52:20 GMT Reproduction and quotation must name Benchize, link to the canonical HTML URL, and retain individual source citations. A single measure of AI's potential economic impact — agentic model performance across finance, coding, legal, and tax tasks, weighted by each sector's share of U.S. GDP. Composite indexes · Models tested: 14 · Score publishers: Vals AI ## Model results | # | Model | Score | ± | Cost | Latency | | | --- | --- | --- | --- | --- | --- | --- | | 1 | [Gemini 4 Argon](https://benchize.com/models/gemini-4-argon-high.md)High | 54.81% | SE ±3.46 pp | $6.45 | 1,881.05source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 54.808 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 6.44908 \| USD \| per_test \| Lower is better \| \| `latency` \| 1881.045 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 262144 provider Google reasoning Not reported Reasoning effort high Temperature 1 topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.46 pp sourcePopulation 23 stderr 3.459 tokenTotals {"cache_read_tokens":739668768,"cache_write_tokens":0,"input_tokens":915877177,"output_tokens":24432059,"reasoning_tokens":11886477} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label google/gemini-4-argon Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/google~1gemini-4-argon` | | | | | | | | 2 | [Muse Wonder](https://benchize.com/models/muse-wonder-max.md)Max | 50.48% | SE ±3.48 pp | $3.98 | 927.11source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 50.481 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 3.980242 \| USD \| per_test \| Lower is better \| \| `latency` \| 927.111 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 131072 provider Meta reasoning Not reported Reasoning effort max Temperature 1 topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.48 pp sourcePopulation 23 stderr 3.475 tokenTotals {"cache_read_tokens":644062808,"cache_write_tokens":0,"input_tokens":732491974,"output_tokens":17268053,"reasoning_tokens":11668590} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label meta/muse-wonder Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/meta~1muse-wonder` | | | | | | | | 2 | [Claude Opus 5.5](https://benchize.com/models/claude-opus-5-5-max.md)Max | 50.48% | SE ±3.48 pp | $25.16 | 6,749.53source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 50.481 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 25.163725 \| USD \| per_test \| Lower is better \| \| `latency` \| 6749.531 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort max harness Not reported maxOutputTokens 128000 provider Anthropic reasoning Not reported Reasoning effort max Temperature 1 topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.48 pp sourcePopulation 23 stderr 3.475 tokenTotals {"cache_read_tokens":1456137656,"cache_write_tokens":254987053,"input_tokens":1711216726,"output_tokens":144935934,"reasoning_tokens":126915044} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label anthropic/claude-opus-5-5 Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/anthropic~1claude-opus-5-5` | | | | | | | | 4 | [Claude Sonnet 5.5](https://benchize.com/models/claude-sonnet-5-5-max.md)Max | 48.08% | SE ±3.47 pp | $14.83 | 5,588.99source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 48.077 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 14.826571 \| USD \| per_test \| Lower is better \| \| `latency` \| 5588.989 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort max harness Not reported maxOutputTokens 128000 provider Anthropic reasoning Not reported Reasoning effort max Temperature 1 topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.47 pp sourcePopulation 23 stderr 3.473 tokenTotals {"cache_read_tokens":2338418360,"cache_write_tokens":311466299,"input_tokens":2650038387,"output_tokens":173348390,"reasoning_tokens":151205430} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label anthropic/claude-sonnet-5-5 Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/anthropic~1claude-sonnet-5-5` | | | | | | | | 5 | [MiMo-V2.6-Pro](https://benchize.com/models/mimo-v2-6-pro.md) | 47.12% | SE ±3.47 pp | $0.18 | 1,816.32source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 47.115 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.181201 \| USD \| per_test \| Lower is better \| \| `latency` \| 1816.322 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 128000 provider Xiaomi reasoning Not reported Reasoning effort unknown Temperature 1 topP 0.95 verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.47 pp sourcePopulation 23 stderr 3.469 tokenTotals {"cache_read_tokens":204771712,"cache_write_tokens":0,"input_tokens":247379616,"output_tokens":20743987,"reasoning_tokens":15408757} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label xiaomi/mimo-v2.6-pro Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/xiaomi~1mimo-v2.6-pro` | | | | | | | | 6 | [DeepSeek V4.1 Flash 1017](https://benchize.com/models/deepseek-v4-1-flash-1017-max.md)Max | 45.19% | SE ±3.46 pp | $0.7 | 2,106.88source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 45.192 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.69856 \| USD \| per_test \| Lower is better \| \| `latency` \| 2106.881 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 384000 provider DeepSeek reasoning Not reported Reasoning effort max Temperature 1 topP 1 verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.46 pp sourcePopulation 23 stderr 3.459 tokenTotals {"cache_read_tokens":1909100928,"cache_write_tokens":0,"input_tokens":2003160078,"output_tokens":87441264,"reasoning_tokens":76652436} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label deepseek/deepseek-v4.1-flash-1017 Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/deepseek~1deepseek-v4.1-flash-1017` | | | | | | | | 7 | [Claude Haiku 5.5](https://benchize.com/models/claude-haiku-5-5-max.md)Max | 43.27% | SE ±3.44 pp | $1.63 | 4,911.73source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 43.269 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 1.633459 \| USD \| per_test \| Lower is better \| \| `latency` \| 4911.734 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort max harness Not reported maxOutputTokens 128000 provider Anthropic reasoning Not reported Reasoning effort max Temperature Not reported topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.44 pp sourcePopulation 23 stderr 3.444 tokenTotals {"cache_read_tokens":1396606112,"cache_write_tokens":248419597,"input_tokens":1645146593,"output_tokens":246336648,"reasoning_tokens":230649495} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label anthropic/claude-haiku-5-5 Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/anthropic~1claude-haiku-5-5` | | | | | | | | 8 | [DeepSeek V4 Pro 0813](https://benchize.com/models/deepseek-v4-pro-0813-max.md)Max | 40.87% | SE ±3.42 pp | $1.09 | 2,527.18source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 40.865 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 1.088017 \| USD \| per_test \| Lower is better \| \| `latency` \| 2527.177 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 384000 provider DeepSeek reasoning Not reported Reasoning effort max Temperature Not reported topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.42 pp sourcePopulation 23 stderr 3.417 tokenTotals {"cache_read_tokens":380911360,"cache_write_tokens":0,"input_tokens":433434003,"output_tokens":35408474,"reasoning_tokens":30213976} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label deepseek/deepseek-v4-pro-0813 Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/deepseek~1deepseek-v4-pro-0813` | | | | | | | | 9 | [GPT-6 Astra](https://benchize.com/models/gpt-6-astra-max.md)Max | 39.42% | SE ±3.4 pp | $10.58 | 1,491.34source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 39.423 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 10.577143 \| USD \| per_test \| Lower is better \| \| `latency` \| 1491.339 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 128000 provider OpenAI reasoning Not reported Reasoning effort max Temperature Not reported topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.4 pp sourcePopulation 23 stderr 3.397 tokenTotals {"cache_read_tokens":215071044,"cache_write_tokens":0,"input_tokens":290661092,"output_tokens":21448775,"reasoning_tokens":16238030} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label openai/gpt-6-astra Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/openai~1gpt-6-astra` | | | | | | | | 10 | [Gemini 3.8 Flash](https://benchize.com/models/gemini-3-8-flash-high.md)High | 38.94% | SE ±3.39 pp | $1.63 | 324.32source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 38.942 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 1.630498 \| USD \| per_test \| Lower is better \| \| `latency` \| 324.318 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 65536 provider Google reasoning Not reported Reasoning effort high Temperature 1 topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.39 pp sourcePopulation 23 stderr 3.389 tokenTotals {"cache_read_tokens":474745174,"cache_write_tokens":0,"input_tokens":581263116,"output_tokens":14420656,"reasoning_tokens":10782340} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label google/gemini-3.8-flash Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/google~1gemini-3.8-flash` | | | | | | | | 11 | [GPT-6.1 Sol](https://benchize.com/models/gpt-6-1-sol-max.md)Max | 38.46% | SE ±3.38 pp | $3.38 | 2,060.48source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 38.462 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 3.384561 \| USD \| per_test \| Lower is better \| \| `latency` \| 2060.48 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 128000 provider OpenAI reasoning Not reported Reasoning effort max Temperature Not reported topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.38 pp sourcePopulation 23 stderr 3.381 tokenTotals {"cache_read_tokens":260063643,"cache_write_tokens":68892844,"input_tokens":329856713,"output_tokens":20763888,"reasoning_tokens":15434407} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label openai/gpt-6.1-sol Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/openai~1gpt-6.1-sol` | | | | | | | | 12 | [MiMo-V2.6-Flash](https://benchize.com/models/mimo-v2-6-flash.md) | 37.98% | SE ±3.37 pp | $0.07 | 919.69source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 37.981 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.070424 \| USD \| per_test \| Lower is better \| \| `latency` \| 919.689 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 128000 provider Xiaomi reasoning Not reported Reasoning effort unknown Temperature 1 topP 0.95 verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.37 pp sourcePopulation 23 stderr 3.373 tokenTotals {"cache_read_tokens":310639232,"cache_write_tokens":0,"input_tokens":371466896,"output_tokens":20269368,"reasoning_tokens":10496896} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label xiaomi/mimo-v2.6-flash Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/xiaomi~1mimo-v2.6-flash` | | | | | | | | 13 | [MiniMax M3.1 Flash](https://benchize.com/models/minimax-m3-1-flash-max.md)Max | 34.14% | SE ±3.3 pp | $0.44 | 2,911.98source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 34.135 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.444777 \| USD \| per_test \| Lower is better \| \| `latency` \| 2911.983 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort max harness Not reported maxOutputTokens 262144 provider MiniMax reasoning Not reported Reasoning effort max Temperature 1 topP 0.95 verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.3 pp sourcePopulation 23 stderr 3.296 tokenTotals {"cache_read_tokens":1135801046,"cache_write_tokens":0,"input_tokens":1230685744,"output_tokens":102073269,"reasoning_tokens":90583933} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label minimax/MiniMax-M3.1-flash Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/minimax~1MiniMax-M3.1-flash` | | | | | | | | 14 | [GPT-6 Luna](https://benchize.com/models/gpt-6-luna-max.md)Max | 30.29% | SE ±3.19 pp | $0.44 | 2,547.75source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 30.288 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.440296 \| USD \| per_test \| Lower is better \| \| `latency` \| 2547.747 \| source_duration \| reported_duration \| Lower is better \| ### Configuration computeEffort Not reported harness Not reported maxOutputTokens 128000 provider OpenAI reasoning Not reported Reasoning effort max Temperature Not reported topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.19 pp sourcePopulation 23 stderr 3.194 tokenTotals {"cache_read_tokens":1044290820,"cache_write_tokens":100299736,"input_tokens":1145599233,"output_tokens":60206390,"reasoning_tokens":55010848} ### Source [Original source](https://www.vals.ai/_astro/benchmark_view_vals_index.ye6C1wH2.json) Source Vals AI Original model label openai/gpt-6-luna Source version 2.2 Observed 2026-10-21T16:45:49.430Z Evidence SHA-256 `0e95aee94cd2bf34bc4b43a8f4bb16c7817de138613c21011873fea4e0786b45` Source locator `/tasks/legal_research/openai~1gpt-6-luna` | | | | | | | --- Cite as: Vals Index / legal research — Benchize, https://benchize.com/benchmarks/vals-ai%3Avals_index%3Alegal_research (retrieved 2026-10-21)