~ same prompt, different LLM — every model gets the same brief in Blender, you judge the render ~
Google / Gemini 3.8 Flash / thinking:medium has entered the arena NEW!
LeaderboardGalleryVoteRunsTasks
RankContestantRatingVotesRunsDone %Success %Avg $Avg callsAvg time
1OpenAI / GPT 6 Astra / thinking:mediumcompare →18981310100$0.69—290s
2Google / Gemma 4 31b Itcompare →1636101100100$0.00—71s
3Anthropic / Sonnet 5 / thinking:lowcompare →1414810100$0.12—158s
4DeepSeek / Deepseek V4 Flashcompare →1383710100$0.00—111s
provOpenAI / GPT 5.6 Lunacompare →1446110100$0.03—64s
provGoogle / Gemini 3.5 Flash / thinking:mediumcompare →1223110100$0.22—174s

Model comparison

each point is a model (One-shot, plaster-bust) — hover for exact values
AnthropicDeepSeekGoogleOpenAI unrated
11221341156117801999$0.001$0.004$0.026$0.175$1.18Bradley–Terry ratingOpenAI / GPT 6 Astra / thinking:mediumGoogle / Gemma 4 31b ItAnthropic / Sonnet 5 / thinking:lowDeepSeek / Deepseek V4 FlashOpenAI / GPT 5.6 LunaGoogle / Gemini 3.5 Flash / thinking:mediumMean cost per task (USD, log scale)

By dimension

separate Bradley–Terry fits from lens rounds + chip verdicts
Lighting
provOpenAI / GPT 6 Astra / thinking:medium1500
provAnthropic / Sonnet 5 / thinking:low1500
Composition
provOpenAI / GPT 6 Astra / thinking:medium1595
provAnthropic / Sonnet 5 / thinking:low1405
Modeling
provOpenAI / GPT 6 Astra / thinking:medium1595
provAnthropic / Sonnet 5 / thinking:low1405
Materials
provOpenAI / GPT 6 Astra / thinking:medium1500
provAnthropic / Sonnet 5 / thinking:low1500
Render
provOpenAI / GPT 6 Astra / thinking:medium1595
provAnthropic / Sonnet 5 / thinking:low1405
Adherence
provOpenAI / GPT 6 Astra / thinking:medium1595
provAnthropic / Sonnet 5 / thinking:low1405

Modes compared

each contestant's ratings across run modes (mixed ladder)
ContestantFullOne-shot3D-only
DeepSeek / Deepseek V4 Flash—1315—
Google / Gemini 3.5 Flash / thinking:medium—1141—
Google / Gemma 4 31b It—1509—
OpenAI / GPT 5.6 Luna172013181742
OpenAI / GPT 5.6 Luna / thinking:max1808——
OpenAI / GPT 6 Astra / thinking:medium—1810—
Anthropic / Sonnet 51307——
Anthropic / Sonnet 5 / thinking:low—1332—
Vote A/BGalleryRunsbradley-terry fit over 20 decisive votes
Votes cast on this bench: 001671 · 163 runs · 26 contestants
Best viewed with Netscape Navigator 4.07 at 800×600 · real runs, live ratings