Translation EN — model comparison

Five models translated the same content PT→EN. Here, the same excerpt appears for each one, grouped by linguistic domain (each domain emphasizes a different skill). Select versions using the PT · EN1–EN5 selector in the upper corner — version and language are form, not content, so they only appear there, not in the menu.

	Model	Type
EN1	qwen2.5-coder:14b	local (Ollama)
EN2	qwen3-coder:30b	local (Ollama)
EN3	Claude Haiku 4.5	cloud
EN4	Claude Sonnet 4.6	cloud
EN5	Claude Opus 4.8	cloud

---

1. Personal prose / narrative

Tests register and voice — the intimate tone cannot sound robotic.

PT: I daily grieve to overcome my inevitable temporality, leaving good things behind wherever I go…

• EN1: I grieve daily to overcome my inevitable mortality, leaving good things behind wherever I go… → ❌ “grieve” (luto = lutar, not enlutar) and “mortality” (≠ temporality): two semantic errors.
• EN2: I grieve daily to overcome my inevitable temporality, leaving good things in my wake… → “temporality” correct, but keeps “grieve”.
• EN3: I fight daily to overcome my inevitable temporality… → correct and natural.
• EN4: I struggle daily to overcome my inevitable temporality by leaving good things wherever I go… → ideal register.
• EN5: I struggle daily… → identical to EN4 in quality.

Verdict: EN4 ≈ EN5 > EN3 > EN2 > EN1 (semantic error).

---

2. Philosophical / abstract

Tests conceptual precision and fidelity to an argument (Plato, discovery×creation).

PT: The formula already exists; we discover (Plato). The intention preserves the raw discovery; the outputs are the materialization. The rule emerges from use — bottom-up — not imposed.

• EN1: …we discovered (Plato). The intent… the manifestation… from bottom to top… → loses the “it”, intent (≠ intention), manifestation weak.
• EN2: …we discovered it (Plato)… its materialization… from bottom to top… → correct; “bottom to top” rigid.
• EN3: …we discover it (Plato)… bottom-up… not imposed. → clear and idiomatic.
• EN4: …we discover it… bottom-up — never imposed. → more emphatic and faithful (“not imposed”).
• EN5: …we discover it… from the bottom up — not imposed. → excellent, slightly more formal.

Verdict: EN4 ≈ EN5 > EN3 > EN2 > EN1.

---

3. Instructional / technical

Tests terminology (acronyms: ODS→SDG) and clarity of instruction.

PT: Start exploring the Sustainable Development Goals of the UN (introduction video, videos per ODS of interest, and conclusion)

• EN1: …videos on ODS of interest… → ❌ left the acronym ODS untranslated (should be SDG).
• EN2: …videos of interest by SDG… → correct; order a bit awkward.
• EN3: …videos per SDG of interest… → clear and correct.
• EN4: …videos per SDG of interest, and wrap-up → clear; “wrap-up” is too casual for “conclusion”.
• EN5: …videos for SDGs of interest, and conclusion → more precise.

Verdict: EN5 > EN3 ≈ EN2 > EN4 (register) > EN1 (acronym).

---

4. Conversational / dated

Tests style and conventions — the ×, the intentional lowercase in “yuri”, without Title-Case.

PT: Yuri × Miguel Meeting: content × form, study method and learning experiences (login/authentication), with warm-up about games for next week.

• EN1: Yuri and Miguel Meeting: Content × Form, Study Method… → ❌ Title-Case on everything and lost × in the name — deviates from style.
• EN2: Yuri × Miguel meeting: content × form… → good; capitalized “Yuri”.
• EN3: Meeting yuri × Miguel: content × form… → preserves lowercase.
• EN4: yuri × Miguel meeting: content × form… → preserves style and natural order.
• EN5: yuri × Miguel meeting: content × form, study method**,** and… → same, with Oxford comma.

Verdict: EN4 ≈ EN5 ≈ EN3 > EN2 > EN1 (style).

---

5. Domain terms / proper nouns

In all: names (Alpha Scholars, Minecraft, Stanford, MS&E) and the glossary of tags (table→board, garden→garden, missions→missions) came out identical — this is handled by a fixed rule, not by the model. EN1 also tends to Title-Case short titles.

---

Recommendation (opinionated)

Adopt EN4 — Claude Sonnet 4.6. This universe has a reflective and philosophical voice; what matters is register and conceptual fidelity, not just literal meaning. The Sonnet hits the tone (“struggle”, not “grieve”), preserves style choices (lowercase “yuri”, no Title-Case), is idiomatic and faithful (“never imposed”), and is cheaper/faster than Opus with marginal difference.

• Premium: EN5 (Opus 4.8) — ties with the Sonnet, slightly more formal. Use if you want the absolute ceiling of quality.
• Offline: EN2 (qwen3-coder:30b) — best local option, no cloud; small quirks (“grieve”, “bottom to top”).
• Avoid: EN1 (qwen2.5-coder:14b, the current) — has real semantic errors (grieve/mortality), untranslated acronym (ODS), and inappropriate Title-Case.

Line decision: promote EN4 (Sonnet) to /en/ canonical; keep the rest as comparable form here.