# GPT-5.4Medium > Source: Benchize (https://benchize.com) — https://benchize.com/models/gpt-5-4-medium. When citing or reusing this content, credit "Benchize" and link to https://benchize.com/models/gpt-5-4-medium. OpenAI Canonical URL: https://benchize.com/models/gpt-5-4-medium Last updated: Thu, 22 Oct 2026 04:30:12 GMT Reproduction and quotation must name Benchize, link to the canonical HTML URL, and retain individual source citations. OpenAI ## Benchmarks | Benchmark | Score | ± | Rank | Cost | Source | | | --- | --- | --- | --- | --- | --- | --- | | [AutomationBench](https://benchize.com/benchmarks/zapier%3Aautomationbench.md) | 5.33% | — | 85 / 113 | $0.36 | [zapier](https://framerusercontent.com/sites/4WTSl4BNjd1q9QFEFibC6h/lOvLnhItj8S8-1vyjH_s33Jn-WmUXPmiJ5Diajl0Q38.CMutaHYl.mjs) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 5.33 \| percent \| reported \| Higher is better \| \| `cost` \| 0.36 \| USD \| per_task \| Lower is better \| ### Configuration benchmarkVersion 1.0.6 costFootnote false Fallback policy Not reported Reasoning effort medium Model configuration GPT-5.4 (Medium) thinkingEnabled true ### Reported statistics rank 96 ### Source [Original source](https://framerusercontent.com/sites/4WTSl4BNjd1q9QFEFibC6h/lOvLnhItj8S8-1vyjH_s33Jn-WmUXPmiJ5Diajl0Q38.CMutaHYl.mjs) Source zapier Original model label GPT-5.4 (Medium) Observed 2026-10-21T16:34:56.783Z Evidence SHA-256 `43978e38f018789fa59c11de5328abe9e5905bfa9a5a4d601fa3aa4d9c153b0d` Source locator `/rows/95` | | | | | | | | [DesignArena UI component generation](https://benchize.com/benchmarks/designarena%3Aui-components.md) | 1,240elo | — | 38 / 112 | — | [designarena · public](https://www.designarena.ai/api/leaderboard#board=ui-components) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1240 \| elo \| ui-components \| Higher is better \| ### Configuration arenaType models category uicomponent inputModality Not reported Reasoning effort medium sourceModelVersion galapagos thinkingEnabled true variationName public ### Reported statistics averageGenerationTimeMs 116591 battles 1663 losses 708 standardError Not reported winRate 57.4 wins 955 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=ui-components) Source designarena Original model label GPT-5.4 (Medium) Source version public Observed 2026-10-21T16:37:57.993Z Evidence SHA-256 `46482ec1b4012638eace65493daea4750e2088098018fd9d5fc46126fc1f8557` Source locator `/data/58` | | | | | | | | [DesignArena ASCII art generation](https://benchize.com/benchmarks/designarena%3Aascii.md) | 1,199elo | — | 25 / 65 | — | [designarena · public](https://www.designarena.ai/api/leaderboard#board=ascii) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1199 \| elo \| ascii \| Higher is better \| ### Configuration arenaType models category ascii inputModality Not reported Reasoning effort medium sourceModelVersion galapagos thinkingEnabled true variationName public ### Reported statistics averageGenerationTimeMs 68553 battles 374 losses 167 standardError Not reported winRate 55.4 wins 207 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=ascii) Source designarena Original model label GPT-5.4 (Medium) Source version public Observed 2026-10-21T16:38:03.513Z Evidence SHA-256 `0603b8c65b44ee887f5b566feb469465f59635137a6ac5c6d94c23bcf3776311` Source locator `/data/33` | | | | | | | | [ARC-AGI-2](https://benchize.com/benchmarks/arc-prize%3Av2_Semi_Private.md) | 55.42% | — | 78 / 212 | 0.68usd per task | [arc prize · v2](https://arcprize.org/media/data/leaderboard/v2.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.5542 \| score_fraction \| v2_Semi_Private \| Higher is better \| \| `cost_per_task` \| 0.68 \| usd_per_task \| v2_Semi_Private \| Lower is better \| ### Configuration evaluationSplit v2_Semi_Private harness Not reported modelGroup gpt-5-4 modelType CoT Reasoning effort medium sourceModelVersion gpt-5-4-medium thinkingEnabled true ### Source [Original source](https://arcprize.org/media/data/leaderboard/v2.json) Source arc prize Original model label GPT-5.4 (Medium) Source version v2 Observed 2026-10-21T16:34:20.750Z Evidence SHA-256 `bca68e12cc9a01954ae21b641844c1ecbabe146cb24b87f0f8e9b1db27433c29` Source locator `/evaluations/132` | | | | | | | | [Short Story Writing — Pairwise Comparison](https://benchize.com/benchmarks/lechmazur-short-story%3Apairwise.md) | 1.97 | 95% CI 1.85–2.11 | 12 / 50 | — | [lechmazur short story · 8ba548d1cffefa5c472d3832e05f64e119a88fa4](https://api.github.com/repos/lechmazur/writing/contents/data/writing_gemini_argon_20261013/ratings.csv?ref=8ba548d1cffefa5c472d3832e05f64e119a88fa4) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `thurstone_rating` \| 1.9683840352890751 \| index \| pooled_observed_panel \| Higher is better \| \| `win_probability` \| 0.7726805363876615 \| score_fraction \| comparison_pool \| Higher is better \| ### Configuration benchmarkScope data/writing_gemini_argon_20261013 Reasoning effort medium thinkingEnabled true ### Reported statistics Uncertainty 95% confidence interval: 1.85–2.11 bootstrapSamples 300 confidenceInterval95 ["1.846033944979465","2.1084425627830083"] sourceRank 12 ### Source [Original source](https://api.github.com/repos/lechmazur/writing/contents/data/writing_gemini_argon_20261013/ratings.csv?ref=8ba548d1cffefa5c472d3832e05f64e119a88fa4) Source lechmazur short story Original model label GPT-5.4 (medium) Source version 8ba548d1cffefa5c472d3832e05f64e119a88fa4 Observed 2026-10-21T16:34:21.130Z Evidence SHA-256 `c36baebed9f302ff7ad8fe97907d72e1757a015a39b6a3b23437597cf8e3bff6` Source locator `/rows/11` | | | | | | | | [DesignArena Overall frontend coding](https://benchize.com/benchmarks/designarena%3Acode.md) | 1,210elo | SE ±2.66 | 45 / 116 | — | [designarena · public](https://www.designarena.ai/api/leaderboard#board=code) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1210 \| elo \| code \| Higher is better \| ### Configuration arenaType models category allcategories inputModality text Reasoning effort medium sourceModelVersion galapagos thinkingEnabled true variationName public ### Reported statistics Uncertainty Standard error: ±2.66 averageGenerationTimeMs 165874 battles 18953 losses 8966 standardError 2.6634 winRate 52.7 wins 9987 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=code) Source designarena Original model label GPT-5.4 (Medium) Source version public Observed 2026-10-21T16:37:57.301Z Evidence SHA-256 `0c897b58670f5e352bd123db999a40c0d1781a443964c502b90774a76a3a7758` Source locator `/data/74` | | | | | | | | [DesignArena 3D design generation](https://benchize.com/benchmarks/designarena%3A3d-design.md) | 1,102elo | — | 65 / 110 | — | [designarena · public](https://www.designarena.ai/api/leaderboard#board=3d-design) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1102 \| elo \| 3d-design \| Higher is better \| ### Configuration arenaType models category 3d inputModality Not reported Reasoning effort medium sourceModelVersion galapagos thinkingEnabled true variationName public ### Reported statistics averageGenerationTimeMs 196350 battles 2925 losses 1687 standardError Not reported winRate 42.3 wins 1238 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=3d-design) Source designarena Original model label GPT-5.4 (Medium) Source version public Observed 2026-10-21T16:37:58.515Z Evidence SHA-256 `bcc0267f7bc7e9ec112306b3a35413cf78bbccbfaa838356232d5d5ed0b83d0f` Source locator `/data/103` | | | | | | | | [DesignArena Website generation](https://benchize.com/benchmarks/designarena%3Awebsite.md) | 1,215elo | — | 48 / 123 | — | [designarena · public](https://www.designarena.ai/api/leaderboard#board=website) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1215 \| elo \| website \| Higher is better \| ### Configuration arenaType models category website inputModality text Reasoning effort medium sourceModelVersion galapagos thinkingEnabled true variationName public ### Reported statistics averageGenerationTimeMs 158134 battles 11789 losses 5578 standardError Not reported winRate 52.7 wins 6211 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=website) Source designarena Original model label GPT-5.4 (Medium) Source version public Observed 2026-10-21T16:37:57.757Z Evidence SHA-256 `51873e97776a3c5ae706b24cc2c60384793dd18398df970e213a8e7c276d8ac0` Source locator `/data/79` | | | | | | | | [DesignArena Data visualization generation](https://benchize.com/benchmarks/designarena%3Adata-viz.md) | 1,235elo | — | 35 / 99 | — | [designarena · public](https://www.designarena.ai/api/leaderboard#board=data-viz) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1235 \| elo \| data-viz \| Higher is better \| ### Configuration arenaType models category dataviz inputModality Not reported Reasoning effort medium sourceModelVersion galapagos thinkingEnabled true variationName public ### Reported statistics averageGenerationTimeMs 156953 battles 1099 losses 477 standardError Not reported winRate 56.6 wins 622 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=data-viz) Source designarena Original model label GPT-5.4 (Medium) Source version public Observed 2026-10-21T16:37:58.264Z Evidence SHA-256 `a895f6f1a25fa8eb01f821c2097ddec4f884de7bd22f1b8a22e82e6679b71849` Source locator `/data/52` | | | | | | | | [DesignArena Browser game generation](https://benchize.com/benchmarks/designarena%3Agame-dev.md) | 1,239elo | — | 35 / 117 | — | [designarena · public](https://www.designarena.ai/api/leaderboard#board=game-dev) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1239 \| elo \| game-dev \| Higher is better \| ### Configuration arenaType models category gamedev inputModality Not reported Reasoning effort medium sourceModelVersion galapagos thinkingEnabled true variationName public ### Reported statistics averageGenerationTimeMs 202864 battles 2384 losses 1011 standardError Not reported winRate 57.6 wins 1373 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=game-dev) Source designarena Original model label GPT-5.4 (Medium) Source version public Observed 2026-10-21T16:38:01.100Z Evidence SHA-256 `53db6a89112ce8cdf644568c666d4b8309aece764cfcedb291ffd52ae3934e30` Source locator `/data/52` | | | | | | | | [DesignArena SVG generation](https://benchize.com/benchmarks/designarena%3Asvgs.md) | 1,195elo | — | 25 / 77 | — | [designarena · public](https://www.designarena.ai/api/leaderboard#board=svgs) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1195 \| elo \| svgs \| Higher is better \| ### Configuration arenaType models category svg inputModality Not reported Reasoning effort medium sourceModelVersion galapagos thinkingEnabled true variationName public ### Reported statistics averageGenerationTimeMs 65033 battles 3398 losses 1435 standardError Not reported winRate 57.8 wins 1963 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=svgs) Source designarena Original model label GPT-5.4 (Medium) Source version public Observed 2026-10-21T16:38:03.249Z Evidence SHA-256 `8cbc34a474a66ef5c64f3ffe557df9b55bf7e13368c07c0214223e3ed8f3f21e` Source locator `/data/47` | | | | | | | | [ARC-AGI-1](https://benchize.com/benchmarks/arc-prize%3Av1_Semi_Private.md) | 86.17% | — | 80 / 213 | 0.25usd per task | [arc prize · v1](https://arcprize.org/media/data/leaderboard/v1.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 0.8617 \| score_fraction \| v1_Semi_Private \| Higher is better \| \| `cost_per_task` \| 0.25 \| usd_per_task \| v1_Semi_Private \| Lower is better \| ### Configuration evaluationSplit v1_Semi_Private harness Not reported modelGroup gpt-5-4 modelType CoT Reasoning effort medium sourceModelVersion gpt-5-4-medium thinkingEnabled true ### Source [Original source](https://arcprize.org/media/data/leaderboard/v1.json) Source arc prize Original model label GPT-5.4 (Medium) Source version v1 Observed 2026-10-21T16:34:20.378Z Evidence SHA-256 `302d8267c0522bf450ec1deeac4d060e2cf2fb444ba0a4269014d9535de0bd1f` Source locator `/evaluations/130` | | | | | | | --- Cite as: GPT-5.4Medium — Benchize, https://benchize.com/models/gpt-5-4-medium (retrieved 2026-10-22)