Loading the page.
hub · LLM ranking for code · live
Models: who delivers more score per dollar
Each model - a line across effort levels: where it sits on the price-quality map and which config is more efficient. Figures - independent DeepSWE, 24h cron snapshot.
GLM 5.3: API, open weights and comparison with Flash
Gemini 4 Argon: access, pricing and first benchmarks
Jev: a model without text, solutions onlyUnion Alpha: a free model in examples
GPT-6.1 Sol: use cases, pricing and comparison with AstraGPT-6 Astra: model card
Claude Fable 5.1: model cardClaude Opus 5.5: Fable tier is cheaperClaude Sonnet 5.5: nearly Opus at half the priceModels for images
GPT Image 2.5: Flare and Sunburst in examplesModels for music
Suno v6: listen to examples and compare versionsVoiceover models
Eleven v4: first in the blind voice rankingupdated October 6, 2026
live · 24h cron snapshot · pass@1 and task price - deepswe.datacurve.ai · lines - effort levels of one model
top 10 · market score · as of Sep 26, 2026
Claude Opus 5.5
Anthropic · Low
100
Claude Sonnet 5.5
Anthropic · Low
96
GPT-6 Astra
OpenAI · Medium
94
Claude Fable 5.1
Anthropic · Low
91
Claude Opus 5
Anthropic · High
80
Claude Fable 5
Anthropic · High
78
Gemini 3.8 Flash
Google · Medium
75
GPT-6 Sol
OpenAI · Low
74
GLM-5.3
Z.ai · Medium
67
GPT-5.6
OpenAI · High
65
forms - vizcn (vibecoding.tech/vizcn) · frontier live from the prod warehouse · top 10 - market score, as of Sep 26, 2026