September 23, 2026
Four Models Shift the Price-Quality Frontier: 11 New Points in the Rankings
Claude Opus 5.5 topped the Intelligence Index on September 23: 58 points at $5,98 per task.
Artificial Analysis
@artificialanlys
This week, the release of MiMo-V2.6-Pro, Claude Opus 5.5, GPT-6 Luna, and GPT-6 Sol shifted the Pareto frontier between the Intelligence Index and cost per task. Together, these models formed 11 new points on the Pareto frontier through different reasoning levels: five for GPT-6 Luna, one each for MiMo-V2.6-Pro and GPT-6 Sol, and four for Claude Opus 5.5. GPT-6 Luna (max) scores 37 points at $0,068 per task, MiMo-V2.6-Pro scores 46 points at $0,13, GPT-6 Sol (max) scores 48 points at $1,06, and Claude Opus 5.5 (max with fallback) became the new highest-scoring model with 58 points at $5,98 per task.
· 8.3K views
Over the week, MiMo-V2.6-Pro, Claude Opus 5.5, GPT-6 Luna, and GPT-6 Sol added 11 points to the price-quality Pareto frontier. Five points came from Luna's different reasoning levels, four from Opus, and one each from MiMo and Sol.
Four points. Artificial Analysis compared the max modes and accounted for fallback for Opus.
| Model | Score | Cost per task | | --- | ---: | ---: | | GPT-6 Luna, max | 37 | $0,068 | | MiMo-V2.6-Pro | 46 | $0,13 | | GPT-6 Sol, max | 48 | $1,06 | | Claude Opus 5.5, max with fallback | 58 | $5,98 |
The cost per task in this index is not the same as the vendor's pricing. Artificial Analysis accounts for input tokens, cache hits and writes, reasoning, and model output. In the Opus run, the model generated 260 million tokens.
API pricing. OpenAI halved prices versus GPT-5.6: GPT-6 Sol costs $2/$10 per 1 million input and output tokens instead of $4/$20, while Luna costs $0,10/$0,50 instead of $0,20/$1,20. For Sol, the index's cost per task ranges from $0,13 on low to $1,06 on max.
Claude Opus 5.5 costs $4/$20 per 1 million tokens, versus $5/$25 for Opus 5. In an early test, the model fixed and verified a 200 000-line codebase in under 3 hours, while Opus 5 took more than 20 hours and 2,5x more tokens.
Where to get the models
GPT-6 Sol and GPT-6 Luna are available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu. In the API, they are called `gpt-6-sol` and `gpt-6-luna`.
Claude Opus 5.5 is called `claude-opus-5-5` in the API and is also available through AWS, Google Cloud, and Microsoft Azure.
Xiaomi is releasing MiMo-V2.6-Pro with open weights under the MIT license. The model has 1T parameters, 42B of them active. The API requires a key from the Xiaomi console, and the model is called through the OpenAI-compatible `base_url=https://api.xiaomimimo.com/v1`.
The four releases already offer 11 modes, each occupying its own point between price and quality.
