1
GPT-6 Astra
- Vendor
- OpenAI
- Score
- 59.6
- Context
- 1.1M
- Pricing (in/out per 1M)
- $10 / $50

Composite ranking of AI LLMs as of 2026-09-18, synthesized from llm-stats.com, Vellum, and LMArena. Includes score, pricing, and context window.
As of 2026-09-18, OpenAI GPT-6 Astra (released Sep 3) tops the composite at 59.6. Anthropic Claude Fable 5.1 (Sep 1) rises to second, and Claude Mythos Preview ties for fourth with the highest reasoning score (55.8). The open-source camp is strong: seven open models including Kimi K3, GLM-5.3, and DeepSeek-V4.1-Flash make the top 15. Ranks 16-20 are GPT-5.6 Terra, Gemini 3.8 Flash (Sep 2, fastest output at 517 tok/s), GLM-5.3-Flash, Muse Spark 1.1, and Qwen3.8-Flash-Next. All 20 ranks are ordered by the llm-stats composite score on a single unified scale.
| Rank | Model | Vendor | Score | Context | Pricing (in/out per 1M) |
|---|---|---|---|---|---|
| 1 | GPT-6 Astra | OpenAI | 59.6 | 1.1M | $10 / $50 |
| 2 | Claude Fable 5.1 |
55.7 |
| 1M |
| $10 / $10 |
| 3 | GPT-5.6 Sol | OpenAI | 54.9 | 1.1M | $6.19 |
| 4 | Claude Mythos Preview | Anthropic | 54.9 | — | — |
| 5 | Claude Opus 5 | Anthropic | 54.6 | 1M | $5 / $25 |
| 6 | Muse Spark 1.3 | Meta | 54.3 | 1M | $1.39 |
| 7 | Claude Fable 5 | Anthropic | 54.3 | 1M | $10 / $50 |
| 8 | Kimi K3 | Moonshot AI | 53 | 1M | $3.39 |
| 9 | GLM-5.3 | Zhipu AI | 52.7 | 1M | $1.33 |
| 10 | Atria Dawn Preview | Shanghai AI Laboratory | 52.3 | — | — |
| 11 | Qwen3.8 Max | Alibaba (Qwen) | 52 | 1M | $1.81 |
| 12 | DeepSeek-V4-Pro-0813 | DeepSeek | 52 | 1M | $0.46 |
| 13 | DeepSeek-V4.1-Flash | DeepSeek | 51.8 | 1M | $0.24 |
| 14 | Hy4 preview | Tencent | 51 | — | — |
| 15 | Claude Opus 4.8 | Anthropic | 50.8 | 1M | $5 / $25 |
| 16 | GPT-5.6 Terra | OpenAI | 50.8 | 1.1M | $2.48 |
| 17 | Gemini 3.8 Flash | 50.5 | 1M | $0.89 |
| 18 | GLM-5.3-Flash | Zhipu AI | 50.2 | 1M | $0.17 |
| 19 | Muse Spark 1.1 | Meta | 49.7 | 1M | $1.39 |
| 20 | Qwen3.8-Flash-Next | Alibaba (Qwen) | 49.1 | — | — |
Comments