Rumor: DeepSeek V4 Imminent, Rivals Opus 4.6 at Lower Price
teortaxesTex · x · 2026-07-18
Rumors suggest DeepSeek V4 is about to transition from preview to official release. According to the cited leaks, V4 Pro scores 80.6% on SWE-bench, closely trailing Opus 4.6 at 80.8%, while costing only $3.48/M for output—a seventh of the latter's price—and supporting a 1M context window. However, the original poster expresses strong skepticism about the leak's authenticity, pointing out that the cited price is actually from April's V4 preview, and Opus 4.6 is also a six-month-old model, suggesting the data has been cobbled together and is misleading.
More from Models
- Is DeepSeek's rumored K3 a scaled-down model, or something bigger? X users debate — teortaxesTex · 2026-09-11
- DeepSeek update keeps cache hits mid-conversation, cuts costs 36.6% — teortaxesTex · 2026-09-11
- AI Sextet offers 6 models free and unlimited for 14 days, including DeepSeek and Qwen — airesearch12 · 2026-09-11
- BullshitBench update: GPT-6-Astra beats all prior OpenAI models but still trails Anthropic — scaling01 · 2026-09-11
- Astra Scores 83% on GauntletBench, First Computer-Use Agent to Beat Human Baseline — ducha_aiki · 2026-09-11
- Kimi K2.8 Preview rolls out: near-K3 coding performance, 1M context for all tiers — teortaxesTex · 2026-09-11