# Vals Index / legal agent benchmark > Source: Benchize (https://benchize.com) — https://benchize.com/benchmarks/vals-ai%3Avals_index%3Alegal_agent_benchmark. When citing or reusing this content, credit "Benchize" and link to https://benchize.com/benchmarks/vals-ai%3Avals_index%3Alegal_agent_benchmark. 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_agent_benchmark 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 | 27.08% | SE ±3.74 pp | $10.55 | 1,755.4source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 27.083 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 10.54836 \| USD \| per_test \| Lower is better \| \| `latency` \| 1755.397 \| 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 Not reported topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±3.74 pp sourcePopulation 23 stderr 3.738 tokenTotals {"cache_read_tokens":733823568,"cache_write_tokens":null,"input_tokens":812145091,"output_tokens":18424207,"reasoning_tokens":5211457} ### 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_agent_benchmark/google~1gemini-4-argon` | | | | | | | | 2 | [Muse Wonder](https://benchize.com/models/muse-wonder-max.md)Max | 23.75% | SE ±3.65 pp | $7.08 | 906.14source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 23.75 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 7.081987 \| USD \| per_test \| Lower is better \| \| `latency` \| 906.14 \| 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: ±3.65 pp sourcePopulation 23 stderr 3.646 tokenTotals {"cache_read_tokens":1374979944,"cache_write_tokens":0,"input_tokens":1414833931,"output_tokens":12111113,"reasoning_tokens":1875186} ### 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_agent_benchmark/meta~1muse-wonder` | | | | | | | | 3 | [DeepSeek V4.1 Flash 1017](https://benchize.com/models/deepseek-v4-1-flash-1017-max.md)Max | 14.17% | SE ±2.79 pp | $0.61 | 1,328.11source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 14.167 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.612596 \| USD \| per_test \| Lower is better \| \| `latency` \| 1328.108 \| 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: ±2.79 pp sourcePopulation 23 stderr 2.787 tokenTotals {"cache_read_tokens":5492827392,"cache_write_tokens":null,"input_tokens":5524481146,"output_tokens":25881985,"reasoning_tokens":10520362} ### 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_agent_benchmark/deepseek~1deepseek-v4.1-flash-1017` | | | | | | | | 4 | [MiMo-V2.6-Pro](https://benchize.com/models/mimo-v2-6-pro.md) | 13.75% | SE ±2.84 pp | $0.22 | 1,861.85source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 13.75 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.216614 \| USD \| per_test \| Lower is better \| \| `latency` \| 1861.848 \| 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: ±2.84 pp sourcePopulation 23 stderr 2.836 tokenTotals {"cache_read_tokens":604899072,"cache_write_tokens":null,"input_tokens":631569981,"output_tokens":10405619,"reasoning_tokens":4622671} ### 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_agent_benchmark/xiaomi~1mimo-v2.6-pro` | | | | | | | | 5 | [MiMo-V2.6-Flash](https://benchize.com/models/mimo-v2-6-flash.md) | 13.33% | SE ±2.82 pp | $0.09 | 1,342.56source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 13.333 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.085646 \| USD \| per_test \| Lower is better \| \| `latency` \| 1342.563 \| 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: ±2.82 pp sourcePopulation 23 stderr 2.821 tokenTotals {"cache_read_tokens":925243136,"cache_write_tokens":null,"input_tokens":958015494,"output_tokens":12423312,"reasoning_tokens":6316080} ### 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_agent_benchmark/xiaomi~1mimo-v2.6-flash` | | | | | | | | 6 | [Gemini 3.8 Flash](https://benchize.com/models/gemini-3-8-flash-high.md)High | 12.08% | SE ±2.58 pp | $3.66 | 1,161.72source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 12.083 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 3.656207 \| USD \| per_test \| Lower is better \| \| `latency` \| 1161.715 \| 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 Not reported topP Not reported verbosity Not reported ### Reported statistics Uncertainty Standard error: ±2.58 pp sourcePopulation 23 stderr 2.579 tokenTotals {"cache_read_tokens":1232055515,"cache_write_tokens":null,"input_tokens":1288268290,"output_tokens":16504243,"reasoning_tokens":6011153} ### 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_agent_benchmark/google~1gemini-3.8-flash` | | | | | | | | 7 | [DeepSeek V4 Pro 0813](https://benchize.com/models/deepseek-v4-pro-0813-max.md)Max | 10.83% | SE ±2.72 pp | $0.8 | 655.41source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 10.833 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.802111 \| USD \| per_test \| Lower is better \| \| `latency` \| 655.413 \| 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: ±2.72 pp sourcePopulation 23 stderr 2.723 tokenTotals {"cache_read_tokens":726363264,"cache_write_tokens":null,"input_tokens":746916941,"output_tokens":9384470,"reasoning_tokens":4362757} ### 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_agent_benchmark/deepseek~1deepseek-v4-pro-0813` | | | | | | | | 8 | [Claude Sonnet 5.5](https://benchize.com/models/claude-sonnet-5-5-max.md)Max | 10.42% | SE ±2.57 pp | $16.61 | 4,397.95source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 10.417 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 16.612583 \| USD \| per_test \| Lower is better \| \| `latency` \| 4397.953 \| 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: ±2.57 pp sourcePopulation 23 stderr 2.572 tokenTotals {"cache_read_tokens":7328811056,"cache_write_tokens":271040714,"input_tokens":7599893900,"output_tokens":71504693,"reasoning_tokens":47513756} ### 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_agent_benchmark/anthropic~1claude-sonnet-5-5` | | | | | | | | 9 | [Claude Opus 5.5](https://benchize.com/models/claude-opus-5-5-max.md)Max | 9.17% | SE ±2.37 pp | $21.38 | 3,507.43source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 9.167 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 21.37949 \| USD \| per_test \| Lower is better \| \| `latency` \| 3507.428 \| 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: ±2.37 pp sourcePopulation 23 stderr 2.366 tokenTotals {"cache_read_tokens":3547556427,"cache_write_tokens":202035592,"input_tokens":3749613119,"output_tokens":44884808,"reasoning_tokens":27684473} ### 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_agent_benchmark/anthropic~1claude-opus-5-5` | | | | | | | | 10 | [MiniMax M3.1 Flash](https://benchize.com/models/minimax-m3-1-flash-max.md)Max | 7.5% | SE ±2.27 pp | $0.47 | 1,497.6source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 7.5 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.472848 \| USD \| per_test \| Lower is better \| \| `latency` \| 1497.601 \| 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: ±2.27 pp sourcePopulation 23 stderr 2.265 tokenTotals {"cache_read_tokens":2703378118,"cache_write_tokens":0,"input_tokens":2730849267,"output_tokens":20117381,"reasoning_tokens":6681765} ### 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_agent_benchmark/minimax~1MiniMax-M3.1-flash` | | | | | | | | 11 | [GPT-6 Astra](https://benchize.com/models/gpt-6-astra-max.md)Max | 7.08% | SE ±1.9 pp | $25.46 | 1,953.36source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 7.083 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 25.460295 \| USD \| per_test \| Lower is better \| \| `latency` \| 1953.362 \| 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: ±1.9 pp sourcePopulation 23 stderr 1.899 tokenTotals {"cache_read_tokens":1276652993,"cache_write_tokens":34120712,"input_tokens":1310790488,"output_tokens":10401199,"reasoning_tokens":3452073} ### 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_agent_benchmark/openai~1gpt-6-astra` | | | | | | | | 12 | [GPT-6.1 Sol](https://benchize.com/models/gpt-6-1-sol-max.md)Max | 6.25% | SE ±1.83 pp | $3.54 | 2,425.63source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 6.25 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 3.536302 \| USD \| per_test \| Lower is better \| \| `latency` \| 2425.625 \| 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: ±1.83 pp sourcePopulation 23 stderr 1.83 tokenTotals {"cache_read_tokens":1191414163,"cache_write_tokens":36281197,"input_tokens":1227717408,"output_tokens":11158751,"reasoning_tokens":3901277} ### 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_agent_benchmark/openai~1gpt-6.1-sol` | | | | | | | | 13 | [GPT-6 Luna](https://benchize.com/models/gpt-6-luna-max.md)Max | 4.17% | SE ±1.63 pp | $0.3 | 1,305.49source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 4.167 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 0.298018 \| USD \| per_test \| Lower is better \| \| `latency` \| 1305.488 \| 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: ±1.63 pp sourcePopulation 23 stderr 1.629 tokenTotals {"cache_read_tokens":1277646907,"cache_write_tokens":44806375,"input_tokens":1322468906,"output_tokens":12364131,"reasoning_tokens":6724988} ### 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_agent_benchmark/openai~1gpt-6-luna` | | | | | | | | 13 | [Claude Haiku 5.5](https://benchize.com/models/claude-haiku-5-5-max.md)Max | 4.17% | SE ±1.52 pp | $4.26 | 2,620.8source duration | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `accuracy` \| 4.167 \| percent \| reported \| Higher is better \| \| `cost_per_test` \| 4.264376 \| USD \| per_test \| Lower is better \| \| `latency` \| 2620.795 \| 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: ±1.52 pp sourcePopulation 23 stderr 1.518 tokenTotals {"cache_read_tokens":4716242036,"cache_write_tokens":188617435,"input_tokens":4904885383,"output_tokens":69221549,"reasoning_tokens":47929104} ### 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_agent_benchmark/anthropic~1claude-haiku-5-5` | | | | | | | --- Cite as: Vals Index / legal agent benchmark — Benchize, https://benchize.com/benchmarks/vals-ai%3Avals_index%3Alegal_agent_benchmark (retrieved 2026-10-21)