OpenAI Terra 6.1 in practice: cheap model plus tiered follow-up beats what benchmarks miss
IamNotGorbachev · reddit · 2026-10-29
The author argues OpenAI's Terra 6.1 highlights the economics of AI that benchmarks fail to capture — YMMV is the key. Forums show simultaneous complaints that Terra is pointless, awful, and awesome: the disappointed hoped it would close the gap to Opus 5.5 (via Sol 6.1) and got a "worse" model; others wanted Sol quality at lower cost; some found it good and cheap enough for coding.
Their use case needs a fine intelligence/cost balance across two sequential API calls: Terra does the heavy lifting in one shot, cheaper than Luna (max); the straightforward follow-up goes to the cheapest, limited Luna (low). Benchmarks miss task-level impact — moving a task to a larger model at a low setting, cheaper than a smaller model at max, pays dividends. Long reasoning chains sometimes just compensate for smaller models' weaker stored world knowledge.
Related event: GPT-6.1 Terra Reviews: Cheap but Middling in Writing Benchmarks(3 posts)→
More from Models
- hunch: GPU-free ~20ms Jev-like decisions built on EmbeddingGemma 2 — iamrobotbear · 2026-10-29
- Rumor: DeepSeek's official API will strip cyber capabilities pending regulatory audits — teortaxesTex · 2026-10-29
- Vitalik: AI is hyper-knowledgeable but unwise, giving surface-level readings of new texts — paulnovosad · 2026-10-29
- No fresh local-capable open models lately — is Qwen 27B still the best? — emax · 2026-10-29
- Leaker says next model after Carbon is ready, waiting on OpenAI and Anthropic — iruletheworldmo · 2026-10-29
- Docs users stuck with a 'dumb' Gemini as Google gates SOTA models behind tiers — rdesh26 · 2026-10-29