Model AI terbaik untuk konteks panjang
Peringkat model berdasar kinerja konteks panjang. Diperbarui Agustus 2026. Berdasar data benchmark independen.
| # | Model | Penyedia | Skor | Kecepatan | Harga (output) |
|---|---|---|---|---|---|
| 1 | GPT-6 Astra (xhigh) | OpenAI | 100.0 | 76 t/s | $50/M |
| 2 | GPT-6 Astra | OpenAI | 100.0 | 82 t/s | $50/M |
| 3 | GPT-6 Astra (high) | OpenAI | 100.0 | 90 t/s | $50/M |
| 4 | GPT-6 Astra (medium) | OpenAI | 100.0 | 98 t/s | $50/M |
| 5 | GPT-5.6 Sol | OpenAI | 100.0 | 72 t/s | $30/M |
| 6 | GPT-6 Astra (low) | OpenAI | 100.0 | 110 t/s | $50/M |
| 7 | GPT-5.4 | OpenAI | 100.0 | 60 t/s | $10/M |
| 8 | Claude Fable 5 | Anthropic | 99.2 | 71 t/s | $50/M |
| 9 | Kimi K3 | Moonshot AI | 99.2 | 37 t/s | $15/M |
| 10 | Claude Opus 5 | Anthropic | 99.2 | 60 t/s | $25/M |
| 11 | Qwen3.8 Max | Alibaba | 99.2 | 55 t/s | $2/M |
| 12 | Claude Opus 4.8 | Anthropic | 99.2 | 50 t/s | $15/M |
| 13 | Muse Spark 1.1 | Meta | 99.2 | 208 t/s | $4.25/M |
| 14 | Gemini 3.6 Flash | 99.2 | 233 t/s | $7.5/M | |
| 15 | Claude Opus 4.7 | Anthropic | 99.2 | 80 t/s | $15/M |
| 16 | Qwen3.7 Max | Alibaba | 99.2 | 203 t/s | $7.5/M |
| 17 | Gemini 3.5 Flash | 99.2 | 267 t/s | $9/M | |
| 18 | MiMo V2.5 Pro | Xiaomi | 99.2 | 65 t/s | $0.87/M |
| 19 | DeepSeek V4 Pro | DeepSeek | 99.2 | 66 t/s | $0.87/M |
| 20 | Claude Sonnet 5 | Anthropic | 99.2 | 96 t/s | $10/M |
GPT-6 Astra (xhigh)
Claude Fable 5
Kimi K3
Qwen3.8 Max
Muse Spark 1.1
Gemini 3.6 Flash
MiMo V2.5 Pro
DeepSeek V4 ProTop 5 \u2014 Perbandingan cepat
Metodologi
Peringkat berdasar evaluasi benchmark independen. Skor dihitung dari beberapa benchmark terstandar dan dinormalisasi ke 0–100.