OpenAI

GPT-6 Astra

OpenAI frontier flagship with Critical-tier autonomous cybersecurity, extreme token efficiency, and 72.6% on OSWorld 2.0.

#4publié le 3 sept. 2026·Proprietary·Not disclosed
textimage1.1M ctxreasoning
API providersvs GPT-5.6 SolClassement
Intelligence
86.2 #4
AAQI 61.0
Output speed
82 t/s
TTFT 143 ms · ITL 12.2 ms/tok
Blended price (3:1)
$20 /1M
In $10/M · Out $50/M
Context
1.1M
≈ 2,750 A4 pages · max out 131072

Quality Domain Radar

6-Axis Frontier Evaluation vs Category Average & Flagship

GPT-6 Astra
Claude Mythos Preview (#1)
Average (68)
Coding (67.2)Mathematics (97.6)Hard Science (GPQA) (96)Agentic & Tool Use (72.6)Knowledge (MMLU) (31.5)Vision / Multimodal (98.2)

Radar axes are approximated from category scores where direct benchmark measurements are missing — see Méthodologie.

API deployment

Compare hosts & latency

Throughput, TTFT and cache pricing across Groq, Cerebras, Together, DeepInfra, Fireworks, Bedrock and Azure.

GroqCerebrasTogether AIDeepInfraFireworks AIAWS BedrockAzure AI
View providers matrix →

Scores par Source

Détails du Modèle

Date de Sortie
3 sept. 2026
Paramètres
Not disclosed
Licence
Proprietary
Fenêtre de Contexte
1,100,000 tokens
Prix Entrée
$10/M
Prix Sortie
$50/M
Vitesse Sortie
82 t/s
Temps au Premier Token
143 ms

Estimated monthly bill

Estimated monthly spend
$0.00
Calculating savings vs GPT-4o...

Cite & embed

GitHub Markdown badge
[![LLMPodium Rank](https://llmpodium.com/badge/gpt-6-astra/rank.svg)](https://llmpodium.com/models/gpt-6-astra)
BibTeX
@misc{llmpodium2026gpt6astra,
  title = {Benchmark Analysis and Intelligence Evaluation of GPT-6 Astra},
  author = {LLMPodium Research},
  year = {2026},
  url = {https://llmpodium.com/models/gpt-6-astra}
}

Face à face

More from OpenAI

Modèles alternatifs

Meilleures alternatives à GPT-6 Astra selon les capacités, le prix et les benchmarks.

0 / 4