DeepSeek publiera-t-elle un modèle nommé R2 ou V5 avant le 31 mars 2027 ?
échéance 2027-03-31
confiance moyenne
preuve niveau 4
68 %estimation Asturna
Critère de résolution
Oui si DeepSeek met en ligne sur Hugging Face ou api-docs.deepseek.com un modèle dont le nom officiel contient « R2 » ou « V5 ». Les variantes de V4 (V4 Pro, V4 Flash, V4-Thinking) ne comptent pas.
État de départ vérifié
DeepSeek V4, V4 Pro et V4 Flash ont été publiés le 24 avril 2026. Aucun R2 n'a jamais été publié malgré des rumeurs depuis 2025. DeepSeek a un historique de sauts de version imprévisibles.
Les arguments ci-dessous ont été produits par le modèle en anglais. Ils sont affichés tels quels plutôt que traduits automatiquement, pour que rien sur cette page ne soit un texte que Asturna n'a pas réellement écrit.
Ce qui fait monter
- The reasoning model line (R1) has not been updated since January 2025; a successor (R2) is highly likely within the 27-month window ending March 2027 [Starting State]
- The 11-month window between the V4 release (April 2026) and the deadline is consistent with the 4-7 month gaps seen in earlier versions (V1 to V3) [Starting State]
- DeepSeek is reportedly raising 80 billion RMB to fund R&D and IPO preparations, which would provide massive capital for training next-generation models [0, 1, 2] (Note: This Tier 4 report is not yet corroborated by Tier 1 or 2 sources)
- Heavy recruitment for the elastic computing team suggests a significant scale-up in training infrastructure [4]
Ce qui fait descendre
- The development cycle between V3 and V4 was 16 months, suggesting that a V5 release within 11 months of V4 may be too aggressive if the trend of increasing complexity continues [Starting State]
- Resolution criteria specifically exclude V4 variants (Pro, Flash, Thinking), which are the most likely intermediate releases before a full V5 [Resolution Criterion]
Ce qui ferait le plus bouger ce chiffre
Whether DeepSeek prioritizes the 'R2' reasoning model or the 'V5' general model, as the 11-month window from the previous major general release is relatively tight.
Comment ce chiffre est fabriqué
| Échéance | 2027-03-31 |
| Taux de base | 65% |
| Meilleure preuve | niveau 4 · 4 article(s) utilisé(s) |
| Estimations | 65 · 70 (écart 5 pts) |
| Confiance | moyenne |
| Plafond de preuve | non déclenché |
| Révisée le | 2026-10-08 09:06 |
| Modèle | gemini-3-flash-preview |
Sources consultées
Historique de la probabilité
2026-10-08 09:06 68% 10 article(s) au dossier · gemini-3-flash-preview
- Ce qui fait monter. The reasoning model line (R1) has not been updated since January 2025; a successor (R2) is highly likely within the 27-month window ending March 2027 [Starting State]
- Ce qui fait monter. The 11-month window between the V4 release (April 2026) and the deadline is consistent with the 4-7 month gaps seen in earlier versions (V1 to V3) [Starting State]
- Ce qui fait monter. DeepSeek is reportedly raising 80 billion RMB to fund R&D and IPO preparations, which would provide massive capital for training next-generation models [0, 1, 2] (Note: This Tier 4 report is not yet corroborated by Tier 1 or 2 sources)
- Ce qui fait monter. Heavy recruitment for the elastic computing team suggests a significant scale-up in training infrastructure [4]
- Ce qui fait descendre. The development cycle between V3 and V4 was 16 months, suggesting that a V5 release within 11 months of V4 may be too aggressive if the trend of increasing complexity continues [Starting State]
- Ce qui fait descendre. Resolution criteria specifically exclude V4 variants (Pro, Flash, Thinking), which are the most likely intermediate releases before a full V5 [Resolution Criterion]
- Ce qui ferait le plus bouger ce chiffre. Whether DeepSeek prioritizes the 'R2' reasoning model or the 'V5' general model, as the 11-month window from the previous major general release is relatively tight.
2026-09-24 09:08 39% 0 article(s) au dossier · openai/gpt-oss-20b (Groq)
Raisons non conservées pour cette révision (l'historique détaillé n'est gardé que depuis peu).