GPT-5.6 Sol claims Claude Mythos parity at a third the cost

My notes

Summary

A 29-second hype TikTok claiming OpenAI shipped GPT 5.6 as three simultaneous models. The headline claim: a “GPT Sol” tier matching Anthropic’s “Mythos Preview” on benchmarks at roughly one third the cost, plus a “Terra” tier that matches the prior 5.5 at half the price. All figures are creator-reported product claims, not verified benchmarks.

Key Insight

What was announced/claimed in the clip (treat as reported, not confirmed):

• Three-model launch. GPT 5.6 ships as a family, not a single model - tiers named “Sol” and “Terra” (third tier cut off mid-sentence).
• “Sol” = parity-at-lower-cost play. Claimed to match a competitor’s frontier “Mythos Preview” on benchmarks while costing ~1/3 as much.
• “Terra” = cheaper iteration. Claimed equal to GPT 5.5 in quality but ~2x cheaper.
• Signal, not proof. The transcript is truncated, the tone is hype (“This is huge”), and no benchmark source or pricing page is cited. Model and tier names (“Sol”, “Terra”, “Mythos”) may be the creator’s shorthand rather than official SKUs.

Directional takeaway: the competitive frontier narrative is shifting toward price-per-quality, not just raw capability - vendors racing to deliver last-gen-frontier performance at a fraction of cost.