GPT-5.6 Launches with Three Edge Advantages Over Claude Mythos—Yet Access Is Restricted
OpenAI unveiled the GPT-5.6 series—Sol, Terra, and Luna—featuring max and ultra modes, a tiered pricing model, benchmark scores that surpass Claude Mythos by up to four points, security testing that uses a third of the tokens, and a limited rollout to about twenty government‑approved partners.
OpenAI announced the GPT-5.6 series, comprising three models: Sol (the flagship), Terra, and Luna. Sol introduces two new modes: max , which extends reasoning time, and ultra , which automatically decomposes a task into sub‑agents, runs them in parallel, and aggregates the results.
Pricing per million tokens is tiered:
Sol – $5 input, $30 output, $0.50 cache
Terra – $2.50 input, $15 output, $0.25 cache
Luna – $1.00 input, $6.00 output, $0.10 cache
A Cerebras‑accelerated version is slated for July, promising an inference speed of 750 tokens per second.
Terminal‑Bench 2.1 evaluation shows the following scores:
GPT‑5.6 Sol Ultra: 91.9%
GPT‑5.6 Sol: 88.8%
Claude Mythos 5: 88.0%
GPT‑5.6 Terra: 84.3%
Claude Fable 5: 84.3%
GPT‑5.5: 83.4%
GPT‑5.6 Luna: 82.5%
Claude Opus 4.8: 78.9%
Gemini 3.1 Pro Preview: 70.7%
Sol Ultra leads Claude Mythos 5 by roughly four percentage points, while Terra matches Claude Fable 5’s performance at half the price.
Security testing with ExploitBench indicates that Sol reaches Claude Mythos‑level capability while consuming about one‑third of the tokens. OpenAI rates Sol’s security as “high” (not “critical”). The model can identify browser vulnerabilities and exploit primitives but, under test conditions, cannot autonomously complete a full attack chain.
OpenAI reports spending over 700,000 A100‑equivalent GPU hours on automated red‑team testing to discover cross‑scenario jailbreaks. A built‑in real‑time refusal system pauses suspicious outputs for review by a larger model.
Access is currently limited to roughly 20 partners approved by the U.S. government; ordinary developers and ChatGPT users cannot use the API yet. OpenAI says the rollout will expand in the coming weeks, with more customers added soon.
Commentary notes that the distribution of cutting‑edge AI capabilities is shifting from private companies toward government oversight, and some critics question OpenAI’s reliance on government approval after being outperformed by Claude.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
AI Engineering
Focused on cutting‑edge product and technology information and practical experience sharing in the AI field (large models, MLOps/LLMOps, AI application development, AI infrastructure).
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.