Side-by-side sourced facts
Both columns quote the same reviewed Epoch AI catalog artifact. The LMArena row quotes the separately pinned snapshot published July 12, 2026 and appears only for an exact source model match. Missing values stay missing.
| Field |  Claude Fable 5 |  DeepSeek-V4-Pro |
|---|---|---|
| Developer | Anthropic | DeepSeek |
| Released | June 9, 2026 | April 24, 2026 |
| Domains | Language, Multimodal | Language |
| Tasks | Language modeling/generation, Question answering | Language modeling/generation, Question answering |
| Access | API access | Open weights (unrestricted) |
| Weights | Closed weights | Open weights |
| LMArena rating | 1507.5 (95% interval 1500.0 to 1515.0, 7,856 battles) | 1457.1 (95% interval 1452.6 to 1461.6, 41,800 battles) |
| Context window | Not reported by this source | Not reported by this source |
| Price | Not reported by this source | Not reported by this source |
Common questions
Is Claude Fable 5 better than DeepSeek-V4-Pro?
According to LMArena's snapshot published July 12, 2026, Claude Fable 5 rates higher (1507.5 vs 1457.1) in the style-controlled text category, and the reported 95% intervals do not overlap in that snapshot. That is one source's conversational-preference measurement, not proof of overall superiority.
Which is newer, Claude Fable 5 or DeepSeek-V4-Pro?
According to Epoch AI's model catalog (retrieved July 14, 2026), Claude Fable 5 is newer: it was released June 9, 2026, while DeepSeek-V4-Pro was released April 24, 2026.
Can you self-host Claude Fable 5 or DeepSeek-V4-Pro?
Only DeepSeek-V4-Pro, according to the Epoch records: it ships unrestricted open weights, while Claude Fable 5 is a closed-weight model with API access only.
Evidence boundary
This page declares no winner beyond what a named source's numbers say. Where the sources do not overlap, the page says so plainly. Task quality, price, context window, latency, and reliability are not reported by these sources, so no claim about them appears here. Both models still require hands-on evaluation on your own workload.